Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
The Role of Unemployment Insurance As an Automatic Stabilizer
During a Recession
July 2010
By
Dr. Wayne Vroman The Urban Institute
Project Director
Jacob M. Benus IMPAQ International, LLC
This report was prepared under a subcontract between the Urban Institute and IMPAQ International. The analysis is part of a multi-year project supported by the U.S. Department of Labor, Employment and Training Administration. The analysis used simulations conducted by the Moody’s Economy.com model. The author thanks Jacob Benus of IMPAQ International, Wayne Gordon and Robert Pavosevich of the Employment and Training Administration, and Marisa Di Natale and Robin Heid of Moody’s Economy.com for help and suggestions. Any errors in the report are the sole responsibility of the author. Opinions expressed in the report do not necessarily represent the views of the U.S. Department of Labor, the Urban Institute, IMPAQ International, or Moody’s Economy.com.
The Role of UI as An Automatic Page i July 2010 Stabilizer During a Recession
TABLE OF CONTENTS
Page EXECUTIVE SUMMARY ........................................................................................................ ii CHAPTER 1. UI AS AN AUTOMATIC STABILIZER ................................................................ 1
1.1 Introduction and Summary .................................................................................. 1 1.2 UI in the 2008-2009 Recession ............................................................................. 2 1.3 Recent UI Legislation ............................................................................................ 4 1.4 Earlier Literature .................................................................................................. 6 1.5 Summary .............................................................................................................. 8
CHAPTER 2. KEY UI BEHAVIORAL RELATIONS IN THE STATES ........................................ 10
2.1 Covered Employment ......................................................................................... 10 2.2 UI Tax Rates ........................................................................................................ 13 2.3 Detailed Tax Rates by State and Industry .......................................................... 17 2.4 Regular UI Recipiency Rates ............................................................................... 18 2.5 Extended UI Benefits .......................................................................................... 24 2.6 Regular UI Replacement Rates ........................................................................... 25 2.7 Summary ............................................................................................................ 29
CHAPTER 3. MODELING THE MACROECONOMIC EFFECTS OF UNEMPLOYMENT
INSURANCE ..................................................................................................... 30 3.1 The Economy.com Model .................................................................................. 30 3.2 Model Structure ................................................................................................. 30 3.3 The Simulation Strategy ..................................................................................... 33
CHAPTER 4. SIMULATION RESULTS.................................................................................. 36
4.1 The Steady Growth (No-recession) Simulations ................................................ 37 4.2 Recession Simulations ........................................................................................ 39 4.3 Regular UI Benefits ............................................................................................. 42 4.4 Extended UI Benefits .......................................................................................... 52 4.5 The Effects of UI Taxes ....................................................................................... 56 4.6 The Net Effect of the UI Program ....................................................................... 58 4.7 The Stabilizing Effect of Unemployment Insurance ........................................... 61 4.8 Summary ............................................................................................................ 67
CHAPTER 5. CONCLUSION ................................................................................................ 69 REFERENCES ...................................................................................................................... 71 APPENDIX A. STATE-LEVEL REGRESSIONS ................................................. Appendix A-1 APPENDIX B. COST OF DOING BUSINESS INDEX ......................................... Appendix B-1
The Role of UI as An Automatic Page ii July 2010 Stabilizer During a Recession
LIST OF TABLES
Page Table 1.1. UI Benefits by Program and as a Percent of GDP in Recession Years, 1949 to 2009 ..................................................................................................... 4 Table 2.1. Summary of Regressions - Annual UI Tax Rates, 1960 to 2007 ...................... 16 Table 2.2. Summary of Recipiency Rate Regressions, 1967 to 2007 ............................... 23 Table 2.3. Summary of Replacement Rate Regressions, 1967 to 2007 ........................... 28 Table 4.1. Time Paths of Real GDP and Real Regular UI Benefits, 2007Q1 to 2010Q4 ... 47 Table 4.2. Real GDP and Real UI Benefits in High and Low Recipiency States, 2007Q3 to 2010Q2 .......................................................................................... 51 Table 4.3. Real GDP and Real Extended Benefits, 2008Q1 to 2010Q2 ............................ 55 Table 4.4. The Effect of UI Taxes on Real GDP, 2007Q1 to 2010Q2 ................................. 58 Table 4.5. Net Effect of UI Program on Real GDP, 2007Q1 to 2010Q2 ............................ 60 Table 4.6. Summary: Estimated Stabilizing Effect of Unemployment Insurance on Real GDP ..................................................................................................... 66
LIST OF CHARTS
Page Chart 2.1. Regular UI Recipiency Rates, 1967 to 2008 .................................................... 19 Chart 2.2. Regular UI Recipiency Rates in Six States, 1967 to 2008 ................................ 20 Chart 4.1 Steady Growth, Real GDP Time Paths, 2008 to 2020 ....................................... 38 Chart 4.2. Unemployment Rates in the Recession and No Recession Scenarios,
2007Q1 to 2013Q4 .......................................................................................... 40 Chart 4.3. Real Regular UI Benefits and UI taxes, 2007Q4 to 2013Q4 ............................ 43 Chart 4.4. Three Time Paths of Real GDP, 2007Q4 to 2010Q2 ........................................ 46 Chart 4.5. Real UI Benefits and Taxes, 2007Q4 to 2013Q4 ($billions) ............................ 54 Chart 4.6. Alternative Indices of Real GDP and Employment, .............................................. 2007Q3 to 2010Q4 .......................................................................................... 62
The Role of UI as An Automatic Page iii July 2010 Stabilizer During a Recession
EXECUTIVE SUMMARY Total U.S. Unemployment Insurance (UI) benefit payments increase automatically during
recessionary periods. This increase in UI benefits during recessionary periods cushions
the macro economy from further decline by helping unemployed workers partially
maintain their purchasing power. That is, by partially compensating the unemployed for
the lost earnings, UI benefits help to break the negative cycle of increased unemployment
leading to reduced consumption, which leads to a further reduction in economic activity.
The cyclical response of regular UI benefits during recessions is often enhanced through
legislation. Specifically, during recessions, typically there has been some form of
federally financed UI benefit extension. Thus, the regular UI program together with
federally financed temporary benefit extensions can have a substantial impact in
cushioning the negative effects of recessions on the U.S. economy.
The UI program incorporates three levels (or tiers) of benefits:
1) Regular UI benefits,
2) Temporary (or emergency) federal benefits (EUC), and
3) Federal-State Extended Benefits (EB).
Regular UI benefits are always available with up to 26 weeks of benefits for most eligible
persons. Temporary federal benefits (Emergency Unemployment Compensation or EUC
in the 2008-2009 recession) are paid under conditions set by emergency federal
legislation. Up to 53 weeks of EUC have been available during the present recession.
Federal-State Extended Benefits (EB) are available in periods when unemployment-
related triggers activate the EB program. EB in the present recession has been available
under temporary unemployment rate triggers with full federal financing (as opposed to
50-50 federal-state financing shares of the permanent EB law). Payments from all three
levels contribute to the stabilizing effect of the UI program. While the financing of UI
(i.e., UI payroll taxes) offsets part of the stabilizing effects of UI benefits, the net effect
of the program is to make the economy more stable.
The Role of UI as An Automatic Page iv July 2010 Stabilizer During a Recession
This report examines the performance of each UI program component as an automatic
stabilizer. The analysis relies heavily on macroeconomic simulations generated by the
Moody’s Economy.com econometric model. Our approach traces the path of the
economy with and without each of these components. By comparing paths, we can
measure the effect of the UI program as a whole and by component as an automatic
stabilizer.
In this report, we examine the impact of the UI program in stabilizing the economy
during a deep recession. Rather than simulating an artificial recessionary scenario, we
use the experience of the recent recession (2008-2009) and examine the time path of the
economy with and without the UI program. Our analysis of the stabilizing performance
of the UI program during 2008Q3-2010Q2 yielded the following conclusions:
The regular UI program closed about one-tenth (0.105) of the real gross domestic
product (GDP) shortfall caused by the recession.
Extended benefits closed about one-twelfth (.085) of the real GDP shortfall
caused by the recession.
Because of lags that reflect experience rating, the response of UI taxes was
delayed with little increase in UI taxes occurring in 2009 and 2010. During
2008Q3-2010Q2, increased UI taxes had essentially no effect on real GDP (a gap
closing proportion of -0.007).
Combining all UI components, we find that, overall, the UI program closed 0.183 of the
gap in real GDP caused by the recession. There is reason to believe, however, that for
this particular recession, the UI program provided stronger stabilization of real output
than in many past recessions because extended benefits responded strongly. Multiplier
effects in real GDP were estimated to average 2.0 for regular UI benefits and also 2.0 for
extended benefits.
The Role of UI as An Automatic Page 1 July 2010 Stabilizer During a Recession
CHAPTER 1.
UNEMPLOYMENT INSURANCE AS AN AUTOMATIC STABILIZER
1.1 Introduction and Summary
A primary reason for establishing UI programs was to provide temporary partial
replacement for the loss of earnings occasioned by unemployment. Since loss of income
from a job is often accompanied by decline in household consumption, an increase in
unemployment accompanies declining general economic activity. The UI program, by
partially compensating for lost earnings, helps to break the negative cycle of increased
unemployment leading to reduced consumption, which leads to a further reduction in
economic activity.
The cyclical response of aggregate UI benefit payments to increased unemployment
during recessionary periods cushions the macro economy from negative shocks by
helping to maintain consumer purchasing power. In other words, UI acts as an automatic
stabilizer of real GDP. Benefit payments increase (decrease) automatically in response to
higher (lower) unemployment.
The countercyclical response of UI benefits can also be enhanced through legislation. In
the past, recession-related federal legislation has temporarily extended unemployment
benefits during severe economic downturns. Prior to the present recession, some form of
federally financed benefit extension was enacted in every recession extending back to
1958.
This report examines the performance of UI as an automatic stabilizer of economic
activity. The analysis relies heavily upon simulations made by the econometric model
supported by Economy.com of Moody’s Investor Service (Economy.com). The model
traces alternative time paths of real GDP, employment, unemployment, other macro
variables, and the payment of UI benefits under different assumptions about output and
The Role of UI as An Automatic Page 2 July 2010 Stabilizer During a Recession
inflation. The model used in the analysis has been developed to simulate economic
activity in the individual states. The principal finding of the analysis is that UI plays a
measurable role as an automatic stabilizer of the economy.
This report proceeds as follows: The present chapter provides a brief overview of the
legislative enactments that affect the performance of UI in the present recession. The
chapter then reviews relevant earlier studies of the UI’s stabilizing role. Particular
emphasis is placed upon two earlier analyses whose findings were derived from
simulations with econometric models. Chapter 2 discusses important behavioral relations
that affect the performance of the UI program in individual states. It examines UI
recipiency rates, replacement rates, and the determination of UI taxes. The relationships
discussed and presented in Chapter 2 have all been incorporated into the Economy.com
state model. Chapter 3 briefly describes the structure of the Economy.com model. One
purpose of the chapter is to show how UI benefits and taxes are integrated into the model.
Chapter 4 presents the findings from several simulations. This chapter estimates singly
and in combination the stabilizing effects of regular UI benefits, extended benefits, and
UI taxes. Finally, Chapter 5 summarizes the results and offers concluding comments,
including suggestions for ways to enhance the UI program’s performance as an automatic
stabilizer.
1.2 UI in the 2008-2009 Recession During 2008-2009 the U.S. economy experienced a very serious recession. By the
broadest measure of economic activity, real GDP, the economy shrank during five of the
six calendar quarters after the fourth quarter of 2007 (the start of the recession) through
the second quarter of 2009. The reductions in real output during the fourth quarter of
2008 and the first quarter of 2009, 5.4 percent and 6.4 percent respectively, represented
the worst back-to-back quarterly performance in more than 50 years. Many now refer to
the present downturn as the “great recession”.
The Role of UI as An Automatic Page 3 July 2010 Stabilizer During a Recession
As real output and employment decreased and unemployment increased, cash payments
from state Unemployment Insurance (UI) programs increased sharply. Payments from
regular UI programs (the program that can pay up to 26 weeks of benefits), which had
totaled $32.0 billion in 2007, increased to $42.6 billion (33 percent) in 2008. With
unemployment increasing persistently from May 2008 through the end of 2009, benefit
payouts in the last half of 2008 were 47.5 percent higher than in the last half of 2007.
Larger increases in regular UI benefits occurred in 2009, with the year’s annual total
reaching $79.2 billion. Since July 2008, benefits for those who exhaust their regular UI
entitlements have also been available. The annual total of extended benefits reached $49
billion in 2009. Clearly, UI program benefits have responded strongly to the recession.
Total (regular plus extended) UI benefit payments in 2009 were $128 billion or 0.9
percent of GDP. The highest payout rate between 1947 and 2009 was 1.05 percent of
GDP in 1975 while the third-highest payout rate was 0.82 percent of GDP in 1958.
Table 1.1 summarizes UI benefit payouts in all post-World War II recessions. Annual
payments are shown separately for three levels or “tiers” of UI benefits: Regular UI,
Federal-State Extended Benefits (EB) and Temporary Federal Benefits (Emergency
Unemployment Compensation or EUC in the 2008-2009 recession). For each recession,
the year of highest payouts is identified and payouts are shown in current dollars
(columns [1]-[4]) and as a percent of GDP (columns [6]-[8]).
Programs paying long-term benefits were first active in the recession of 1958 and EB was
first paid in the recession year 1971. The following three observations are drawn from
Table 1.1:
1) Total benefits ranged between 0.49 and 1.01 percent of GDP across the 11
recessionary years (this variation reflects both differing recession severity and
differing availability of long-term benefits).
2) The highest total payout rate occurred in 1975 and the highest payout of
extended benefits (EUC + EB) occurred in 2009.
3) With the addition of 2009 to the table, there is no obvious trend across the 11
recessions (column [8]).
The Role of UI as An Automatic Page 4 July 2010 Stabilizer During a Recession
Table 1.1. UI Benefits by Program and as a Percent of GDP in Recession Years, 1949 to 2009
Recession Year
Regular State
UI
Federal State EB
Temporary Federal Benefits
Total UI
Benefits GDP
Regular Benefits/
GDP
Extended Benefits/
GDP
Total Benefits/
GDP Total [1+2+3] [1]/[5] % [2+3]/5 % [4]/[5] %
[1] [2] [3] [4] [5] [6] [7] [8] 1949 1.7 - - 1.7 266 0.65 - 0.65 1954 2.0 - - 2.0 381 0.53 - 0.53 1958 3.5 - 0.3 3.8 467 0.75 0.06 0.82 1961 3.4 - 0.6 4.0 546 0.63 0.11 0.74 1971 4.9 0.7 0.0 5.6 1,129 0.44 0.06 0.49 1975 11.9 2.5 2.1 16.5 1,635 0.73 0.28 1.01 1980 14.1 1.7 0.0 15.8 2,788 0.51 0.06 0.57 1982 21.3 2.4 1.2 24.9 3,253 0.65 0.11 0.77 1992 24.9 0.0 13.5 38.4 6,342 0.39 0.21 0.60 2002 41.9 0.2 10.7 52.8 10,642 0.39 0.10 0.50 2009 79.2 6.1 43.1 128.4 14,256 0.56 0.35 0.90
Source: Data from U.S. Departments of Labor and Commerce. Data in $billions.
1.3 Recent UI Legislation
The current recession has witnessed a strong policy response intended to help
unemployed workers and their families. In late June 2008, the Congress passed and
President Bush signed the Emergency Unemployment Compensation Act (EUC). This
provided 13 weeks of added benefits to persons who had exhausted their regular UI
benefits. During August and September, the number of EUC claimants exceeded 1.25
million per week, but then the numbers decreased as this added entitlement was also
exhausted. By November, the EUC weekly numbers had declined to about 0.75 million.
During these fall months, the number of regular UI claimants continued a steady ascent,
reaching an average of 4.5 million in December 2008.
EUC was given a second legislative authorization in November 2008. This extended the
period for new EUC claims to the end of March 2009, and increased potential EUC
weeks from 13 to either 20 or 33, depending upon the state’s recent three-month average
The Role of UI as An Automatic Page 5 July 2010 Stabilizer During a Recession
total unemployment rate (TUR). States with a TUR of at least 8.0 percent could pay up to
33 weeks of EUC; other states could pay up to 20 weeks.1
The American Recovery and Reinvestment Act (ARRA) of February 2009 included
several UI provisions. The most important were the following:2
1) The EUC08 program was further extended to December 31, 2009 with unchanged rules for 20 and 33 potential weeks of EUC benefits. New claims for EUC could be received through the end of 2009, with payments extending into 2010 for eligible claimants. A person filing late in 2009 could potentially receive EUC through May 2010.
2) All recipients of UI benefits had their weekly benefit increased by $25 while ARRA provisions were in effect. In a program where the national average weekly benefit was about $300, this represented an 8 percent increase in the overall weekly benefit. The percentage increase was even larger for low-wage claimants and those in low-wage states.
3) The first $2,400 of UI benefits in 2009 was exempted from the federal personal income tax.
4) For UI claimants faced with the loss of health insurance, coverage could be purchased with the federal government paying 65 percent of the monthly premium.
5) The Federal-State Extended Benefits (EB) program was modified to allow easier access to EB payments and longer potential duration (a maximum of 20 weeks in several states rather than the traditional 13). During 2009, more than half the states modified the unemployment rate triggers that activate EB, modifications that will lapse when ARRA lapses.
Both extended benefits programs (EUC and EB) were modified several times during late-
2009-early 2010 to lengthen their availability to the long term unemployed. The most
recent extension allows new claims for EUC through the week of June 2, 2010, and EUC
payments on established claims can occur as late as the week of November 6, 2010.
1 Potential weeks of entitlement to extended benefits is usually expressed as a fraction of the potential weeks of regular UI. Thus the original EUC08 program could pay the lesser of 13 weeks or half of potential duration under the regular UI entitlement. Most states provide for a variable duration of regular UI benefits. Thus, someone entitled to 20 weeks of regular UI would be entitled to only 10 weeks of EUC08. 2 One summary of the UI provisions in ARRA is given in Vroman (2009).
The Role of UI as An Automatic Page 6 July 2010 Stabilizer During a Recession
The net effect of the ARRA has been to substantially increase the total volume of UI
benefit payments in 2009 and 2010. Estimates of the increase in benefit payouts due to
ARRA are necessarily imprecise, since the full depth and duration of the recession are
uncertain. A global estimate of all ARRA provisions affecting benefit payouts would be
at least $60 billion in calendar year 2009. When these are added to payouts under the
regular UI program, the combined total reached $128 billion in 2009. The $128 billion
represented 0.9 percent of GDP in 2009, the second highest percentage over the 63 years
between 1947 and 2009. A similar percentage may occur in 2010.3
1.4 Earlier Literature
A primary objective of UI is to provide built-in or automatic stability to the overall
economy. The economic literature that assesses the strength of UI as an automatic
stabilizer is extensive. For example, Gruber (1997) found that the amount that a family
spends on food falls by 7 percent when the head of the household becomes unemployed;
it would have declined 22 percent in the absence of unemployment benefits.
Two studies of the stabilizing effect of the UI program were supported by the U.S.
Department of Labor. Dunson, et al. (1991) used the Data Resources Incorporated (DRI)
macro model to assess UI’s stabilizing effectiveness. Chemerine, et al. (1999), in an
analysis by Coffey Communications, used the Wharton Economic Forecasting Associates
(WEFA) model.4
Dunson, et al. (1991) and Chimerine, et al. (1999) both conducted broad reviews of
previous literature. The review in Dunson, et al. (1991) described 13 separate studies
using an aggregate income-expenditure approach to assess stabilizing effectiveness.
These studies, published between 1960 and 1986, differed widely in their methodology.
3 The model estimates presented in this paper were based on February 2009 ARRA provisions which were slated to fully expire in May 2010. The model-based analysis did not include effects of the post-ARRA extensions of EUC that were enacted in November 2009, March 2010 and July 2010. The simulated phase-down of 2010Q1 and 2010Q2 were based on the phase-down contemplated under ARRA. 4 The DRI model, the WEFA model, and the model of Chase Econometrics have been combined into the Global Insight macro model, which currently provides forecasting services for several federal agencies, agencies of state government, municipalities, and numerous private businesses.
The Role of UI as An Automatic Page 7 July 2010 Stabilizer During a Recession
All concluded that UI helps to stabilize the overall economy, but the estimates of
stabilizing effectiveness varied quite widely--from reducing real GNP fluctuations by
one-fourth or more (Eilbott 1966), to practically no stabilizing effect. An average
estimate from this set of studies would be that UI prevented roughly 15 percent of the
decline that would have otherwise occurred in aggregate real output. Among the studies
that explicitly considered both UI taxes as well as benefits, most concluded that nearly all
of the stabilizing effect was provided by UI benefits and that UI taxes played either a
small or an inconsistent role.
Dunson, et al. (1991) utilized the DRI model in their simulation analysis. They noted a
downtrend in UI recipiency between the late 1970s and the early 1990s. Their simulations
focused on recession-related changes in real GDP and aggregate employment in the late
1970s and the early 1990s. For both periods, there were two simulations: One with the
UI program operating in its usual manner and one with UI variables frozen in real terms
at levels from the pre-simulation period. The effectiveness of UI was measured during the
four quarters of the largest decrease in real output. In each simulation period, the
percentage difference in real output and employment was measured and averaged. For the
earlier 1970’s period, UI reduced the decline in real GNP by an average of 5.5 percent
and the decline in employment by 4.9 percent. For the latter (forward-looking) period, UI
reduced the decline in real GNP by 3.7 percent and the decline in employment by 3.5
percent. Based on these results, the authors concluded that UI in the 1990s was only 68.5
percent as effective compared to the late 1970s in stabilizing real GNP and 71.4 percent
as effective in stabilizing employment. It should be noted that their results focused upon
just the regular UI program and did not consider extended benefits programs.
The second large-scale model-based analysis was conducted by Chimerine, et al. (1999)
at Coffey Associates. They used the WEFA quarterly econometric model to examine the
performance of UI as an automatic stabilizer over five previous recessions (1970, 1974,
1980, 1982, and 1991). Their principal conclusion was that UI provides substantial
automatic stabilization to the macro economy. They estimated that recession-related
changes in real GDP were reduced on average by about 15 percent by UI benefit
The Role of UI as An Automatic Page 8 July 2010 Stabilizer During a Recession
payments. They also concluded the stabilizing effect of UI on the economy had not
trended downward over their periods of analysis.
In contrast to Dunson, et al., this study focused upon all three tiers of UI benefit
payments (regular UI, temporary federal benefits, and EB). They found (Chapter 5 and
Appendices D and F) that the three tiers of benefit payments had very similar stabilizing
effects per dollar of expenditures. They also documented the decreased scope of the EB
spending after 1981 due to changes in the EB triggers and to a federal bypass option. The
latter allowed states during the 1991 recession to bypass EB and pay temporary federal
benefits to regular UI exhaustees. Nearly all states exercised this option, since it meant
lower EB payments and associated state costs because half of EB is a state fiscal
responsibility, whereas none of EUC is state-funded.
Finding that the need for UI as a stabilizer has not diminished, Chimerine, et al., offered
suggestions for ways to enhance the stabilizing effectiveness of UI. Three changes to
improve effectiveness would be to: 1) raise UI recipiency rates, 2) make the extended
benefit programs more automatic, and 3) increase the level of funding of UI programs.
They also recommend more quantitative analysis of UI with the objective of improving
its performance as an automatic stabilizer. Like the Chimerine, et al. analysis, the present
project will examine the effects of extended benefits as well as regular UI program
benefits.
1.5 Summary
In response to the recession of 2008-2009, federal legislation has increased the scope and
level of UI benefit payments. Federal policy, plus the built-in features of regular UI,
mean that the program will roughly double benefit payouts in 2009 compared to 2008.
Benefit payments in 2009 will be more than triple total payouts in the pre-recession year
2007.
The Role of UI as An Automatic Page 9 July 2010 Stabilizer During a Recession
Previous evaluations of the UI program have found it to be an important automatic
stabilizer of economic activity. These results, however, have not yielded a consensus
estimate of UI’s stabilizing effect. In this report we attempt to improve on previous
studies by conducting a state-level analysis to assess the program’s stabilizing
performance during a severe recession similar to the recession of recent in 2008-2009.
The Role of UI as An Automatic Page 10 July 2010 Stabilizer During a Recession
CHAPTER 2. KEY UI BEHAVIORAL RELATIONS IN THE STATES
The economies of individual states differ in a variety of ways. Contrasts in industrial
structure, money wage levels, demographics (including population growth and labor
force age), and cyclical sensitivity are but a few of the state-specific factors important to
state economic performance. The Economy.com modeling approach incorporates many
state-specific factors into the structure of its state models.5
To simulate the performance of unemployment insurance (UI) as an automatic stabilizer,
it is important to consider state-level differences in economic structures as well as state
differences in UI programs. This chapter focuses on five relationships that characterize
key aspects of the UI programs in the individual states:
1) Determination of covered employment,
2) Average tax rate as a percent of UI covered payroll,
3) Average tax rate by detailed industry within each state,
4) UI recipiency rate (beneficiaries as a proportion of total unemployment) and
5) UI replacement rate (the ratio of the average weekly benefit to the average weekly
wage).
For 2), 4), and 5), regression relationships were developed using annual time series data.
To determine the average tax rate by state and industry, a proportional relationship to the
statewide average tax rate in 2007 was calculated and projected to hold for all future
years spanned by the simulations. The chapter text summarizes these relationships.
(Appendix A displays three sets of state-level regressions.) The relationships yield
accurate estimates of UI benefits and taxes in the individual states.
2.1 Covered Employment
Nearly all employers and wage and salary workers are covered by the UI program. The
only important exceptions are federal government employees, recently discharged service
5 One description of the state models is given in Cochrane (2006). Chapter 3 describes the models.
The Role of UI as An Automatic Page 11 July 2010 Stabilizer During a Recession
members who are covered by separate programs,6
and some employees of small firms
and religious organizations.
Employment covered by UI is of two types: Taxable and reimbursable. Taxable
employers account for more than 80 percent of covered employment. Their UI taxes are
determined by the experience rating system followed in their state. The details of these
systems differ widely, but all set UI taxes in such a way that higher payouts of UI
benefits cause future UI taxes to be higher for most individual employers (all but those
already at the maximum tax rate). Experience rating is described as imperfect, in that
there is not a one-to-one correspondence between changes in UI benefit payouts and
changes in UI taxes for individual employers. Taxes paid by employers flow into state UI
accounts maintained at the U.S. Treasury. These same accounts are the source of benefit
payments to eligible claimants in the regular UI program, that is, the program that can
pay up to 26 weeks of benefits (28 weeks in Montana and 30 weeks in Massachusetts).
The remaining covered employers are reimbursable employers. At the end of each year
they make a payment to the state UI trust fund for all benefits charged to their accounts.
In the aggregate, reimbursable employers account for just under 20 percent of covered
employment. In 2007, for example, reimbursable employment totaled 25.8 million, or
19.3 percent of total covered employment of 133.4 million. Current coverage provisions
have been in place since 1978. Between 1978 and 2007, the reimbursable share of
covered employment increased from 17.6 percent to 19.3 percent.
Two groups of employers have reimbursable coverage: State and local governments and
nonprofit employers. Employment in state and local governments is easily identified, but
nonprofit employment is widely distributed across the industry structure. According to
analysis at the Urban Institute, total nonprofit employment in 2005 was 12.9 million. The
three two-digit industries with the largest amount of nonprofit employment in descending
6 Respectively these are Unemployment Compensation for Federal Employees (UCFE) and Unemployment Compensation for Ex-servicemen (UCX). Payments under these two programs are administered by state UI programs, but they have their own financing that is part of the federal budget. The self-employed also fall outside the scope of UI coverage.
The Role of UI as An Automatic Page 12 July 2010 Stabilizer During a Recession
order of size are: Industry 62 – Health Care and Social Assistance; industry 81 – Other
Services, Except Government; and industry 61 – Educational Services. These three
industries combined accounted for 93.5 percent of nonprofit employment in 2005.7
Nonprofit employment in industry 62 was 7.0 million in 2005 or 54.2 percent of the
nonprofit total. Growth of the nonprofit share of total covered employment undoubtedly
reflects the rapid growth of health sector employment.
Because taxable and reimbursable employers have different UI tax treatment, the state-
level models should distinguish the two types of employers. Following discussions with
staff at the Office of Workforce Security and the Bureau of Labor Statistics, we have
partially addressed this question, but limitations on existing data availability have made it
necessary to follow a methodology where nonprofit employment has been combined with
for-profit private employment. Employment in the government sector (at all levels) was
removed from the total employment estimates. However, when the Bureau of Labor
Statistics publishes state-by-industry data on UI covered employment and payroll,
nonprofit employment is not routinely separated from for-profit employment.8
In
industries with large nonprofit employment, UI-based tax rates will overstate actual tax
rates.
At the level of statewide aggregates, the UI reporting system does distinguish each of
nonprofit employment and government employment from for-profit employment. The
reporting system also records the average contribution rate among for-profit employers.
The Economy.com state models have estimated regressions to determine nonprofit
employment. The regressions use NIPA employment9
7 Industries are classified according to the North American Industrial Classification System (NAICS) codes. See Table 2.2 in Wing, et. al (2008) for 2005 estimates of nonprofit employment by industry.
in the three industries identified
above (NAICS codes 62, 81 and 61) as explanatory variables with different coefficients
8 These data are commonly referred to as Quarterly Census of Employment and Wages (QCEW) 9 NIPA (National Income and Product Accounts) employment is estimated quarterly by the Office of Business Economics in the Commerce Department. The Economy.com models have estimates of NIPA employment by state for detailed industries.
The Role of UI as An Automatic Page 13 July 2010 Stabilizer During a Recession
estimated for the three industries. The CES employment estimate for the state and local
government drives the UI covered employment estimate for this sector.
A regression also determines estimated taxable employment. The explanatory variable for
this regression is total CES employment after removing employment in the federal, state,
and local sectors, and the nonprofit components of employment in sectors 62, 81 and 61.
Total payroll of taxable and of reimbursable employers is also estimated by regression.
The ratio of estimated total payroll to estimated employment is then used in the state
models to estimate average weekly wages for taxable employers, reimbursable
employers, and all employers combined. The estimates of average weekly wages, in turn,
are used in the replacement rate regressions (described below).
Although reimbursable employment accounts for a sizable share of total covered
employment, UI claims against reimbursable employers are typically modest. In 2007, for
example, benefits paid by reimbursable employers totaled $1.7 billion (5.6 percent of
total regular UI benefits). The vast majority of regular UI benefits are paid to current and
former employees of taxable employers, and these benefits are financed by experience-
rated payroll taxes.
2.2 UI Tax Rates
State UI programs use two main methods for setting tax rates for individual taxable
employers. Of the 51 UI programs examined here, 33 use reserve ratio experience rating,
13 use benefit ratio experience rating, two use a combination of reserve ratios and benefit
ratios, three use other systems.10
10 Puerto Rico and the Virgin Islands are not included in this analysis. Michigan and Pennsylvania use both reserve ratios and benefit ratios to set tax rates. Delaware and Oklahoma use benefit-wage ratios, i.e., the wages of employers with benefit charges, while Alaska uses payroll declines to set tax rates.
Reserve ratio systems use the employer fund balance on
a set date (the computation date, most commonly June 30) measured as a percentage of
recent (taxable or total) payrolls to calculate the employer’s reserve ratio. The reserve
ratio then determines where along a schedule of tax rates the employer is located, with
higher tax rates for employers with lower reserve ratios. This tax rate applies throughout
The Role of UI as An Automatic Page 14 July 2010 Stabilizer During a Recession
the entire upcoming year. Benefit ratio states use the benefit payout rate (benefits charged
to an employer as a proportion of the employer’s recent [taxable or total] payroll) to
calculate a benefit ratio, which determines next year’s tax rate. Most states have several
tax rate schedules with higher schedules applicable as the state’s trust fund descends to
lower levels. Higher payouts in both systems (either higher benefit ratios in benefit ratio
systems or lower reserve ratios in reserve ratio systems) cause UI taxes to be higher
automatically in later periods unless overridden by state legislation. The determination of
tax rates for individual employers also depends upon other factors, such as the prevalence
of socialized benefit charges, the turnover rate of covered employers, the minimum tax
rate, the maximum tax rate, and the level of the taxable wage base.
We used regression analysis to examine UI tax rates measured as a percentage of total
payrolls of taxable employers. The regressions showed that lagged benefit ratios exert a
strong positive effect on tax rates while lagged reserve ratios had a negative effect on the
tax rate in most states. However, the explanatory power of lagged benefit ratios was
much higher than for reserve ratios. As a result, we only use lagged benefit ratios in our
analysis.
Table 2.1 displays summary statistics from the regressions (the individual state-level
regressions appear in Table A.1 of Appendix A). Note in Panel A, 41 of 51 regressions
have adjusted R2s of at least 0.60 and the average adjusted R2 is 0.712. The standard
errors are generally small, with all but five smaller than 0.25. The average standard error
of 0.174 is less than 0.20 of the overall tax rate, which averaged 0.940 for the entire set of
2,958 state-year observations.
The benefit ratio slope coefficients in Panel B are nearly all positive, as expected. Of the
204 slopes, 200 are positive and 126 are significant (using a t ratio of 2.0 to denote
significance). The right-hand column in Panel B indicates that the time profile of the
benefit ratio coefficients is quite flat, with the average coefficients ranging between 0.252
(two year lag) and 0.176 (4-year lag). The sum of the four coefficients in Panel B (0.869)
is similar to the median of the sum of the four benefit ratio coefficients in Panel C
The Role of UI as An Automatic Page 15 July 2010 Stabilizer During a Recession
(0.850). Both of these sums are less than 1.0, indicating that using an alternative
specification where the constant term was constrained to 0.0 would have yielded a
coefficient sum even closer to 1.0.
One curious aspect of these regression results is the pattern of the residuals during 2000-
2007. These eight years generate 408 state-year observations. For each state, the size and
sign of each regression residual was noted. If a random process generated the residuals,
one would expect roughly 204 to be positive and 204 to be negative. In fact, there were
only 111 positive residuals compared to 297 negative residuals. The average residual for
these last eight years of the estimation period was negative for 40 of the 51 state
programs, meaning that the predicted tax rates were typically higher than the actual rates.
This raises the question of why effective tax rates were not higher during these years.
This would seem to be a good topic for further research to document state actions that
reduced effective UI tax rates during 2000-2007. The state model uses add factors to
offset the tendency for the regressions to overestimate tax rates in 2009 and later years.
Overall, these results are as expected given the UI program structure and intent. Increases
in the benefit payout rate (benefit ratio) cause the average effective tax rate to change in
the same direction. The vast majority of slope coefficients (98 percent) have the expected
positive signs and the majority (62 percent) is statistically significant. On average, the
regressions indicate the response of the tax rate to changes in benefit payouts is spread
over 4 years, and, in most states, the total response is nearly as large as the change in the
benefit ratio.
The Role of UI as An Automatic Page 16 July 2010 Stabilizer During a Recession
Table 2.1. Summary of Regressions - Annual UI Tax Rates, 1960 to 2007
Panel A. Summary Statistics for 51 Programs
Panel B. Sign and Significance of Coefficients
Positive, Positive, Negative, Negative, Significant Not Signif. Not Signif. Significant Average
Constant 21 12 10 8 0.103 Ben. Ratio Lag 1 Year 32 19 0 0 0.236 Ben. Ratio Lag 2 Years 34 17 0 0 0.252 Ben. Ratio Lag 3 Year 27 24 0 0 0.205 Ben. Ratio Lag 4 Years 33 14 4 0 0.176 Ben. Ratio Sum 0.869
Panel C. Sum of Four Benefit Ratio Coefficients
Number of States Below 0.60 5 0.60-0.699 8 0.70-0.799 10 0.80-0.899 6 0.90-0.999 9 1.00-1.099 5 1.10-Plus 8
Median 0.850
Source: All entries based on 51 state-level regressions in Table A.1 of Appendix A.
Adjusted R2 Standard Error
Number of
States Number of
States Below 0.50 8 Below 0.10 3 0.50-0.599 2 0.10-0.149 20 0.60-0.699 9 0.15-0.199 13 0.70-0.799 13 0.20-0.249 10 0.80-0.899 18 0.25-0.299 3 0.90 Plus 1 0.30 Plus 2 Average 0.712 Average 0.174
The Role of UI as An Automatic Page 17 July 2010 Stabilizer During a Recession
2.3 Detailed Tax Rates by State and Industry
Tax rates on covered employers are known to vary widely across industries within states.
Experience rating of UI taxes ensures that industries with higher benefit payout rates are
subject to higher effective tax rates (taxes as a percent of total covered payroll) than
industries with low payout rates. However, the national UI data reporting system no
longer routinely publishes details on state-level tax rates by industry. The last year of
published data refers to tax rates in 1994.
The QCEW reporting system does record UI contributions in addition to details on
employment, total payroll, and UI taxable payroll. For calendar year 2007, we executed a
tabulation at the state level of contribution rates by industry for private (for-profit plus
nonprofit) employers. The industry detail was at the level of 2-digit NAICS codes, which
span 19 detailed industries. We then divided the industry tax rates by the statewide
average contribution rate to yield a set of 19 relative tax rates for each state.
Individual industries in each state have highly varied claims experiences, which (through
experience rating) cause their tax rates to differ. Industries such as agriculture and
construction, administrative and waste services, and accommodation and food services
have persistently high claims relative to the all-industry average, and their tax rates are
consistently above average. Conversely, low claims volume and associated low tax rates
characterize utilities, finance and insurance, management companies, and health care and
social assistance. In the former industries, average tax rates are frequently twice the all-
industry average, while in the latter group the tax rate often averages less than half the
all-industry average. Relative tax rates within an industry tend to be stable over time for
many industries.
The use of NAICS coding for classifying industries also provides helpful detail on tax
rates within the broad services sector. NAICS codes identify eight broad service sector
industries. For the eight sectors combined, the average tax rate nationwide is only
somewhat below the all-industry average (0.58 percent versus 0.61 percent in 2007),
The Role of UI as An Automatic Page 18 July 2010 Stabilizer During a Recession
Three of the underlying industries have low and three have high average tax rates.
Disaggregation of the services sector provides revealing details about UI tax rate
variation that are not suggested by the average tax rate for the overall service industry.
These relative tax rates can then be multiplied by each statewide average tax rate to yield
estimated tax rates for 19 broad industries. The average tax rates can be obtained using
the tax rate regressions described in the previous section. In simulation results to be
discussed in Chapter 4, the relative tax rates from 2007 were used to estimate industry-
level tax rates for future years. For each future year in a given state, UI tax rates vary by
industry and according to the past 4 years’ experience in paying regular UI benefits.
Because the UI tax rate estimates are based on total payroll, they can be directly entered
into the Economy.com model estimates of the cost of doing business. Employer UI taxes
are one component of labor costs by industry. Thus, within the state models, increases in
UI benefit payouts lead to increases in average UI tax rates. This feedback from benefit
payouts onto UI taxes allows the analysis to estimate the dampening effect of UI taxes on
the performance of UI as an automatic stabilizer of the macro economy.
2.4 Regular UI Recipiency Rates
Only a minority of the unemployed collect regular UI benefits at any point in time. The
recipiency rate as measured here is the ratio of weekly UI beneficiaries (in the regular UI
program or EB) to total unemployment (TU) as measured in the monthly labor force
survey of households. This ratio averaged 0.316 between 1967 and 2007. Readers should
note that this measure of the recipiency rate differs from the measure used by many in
ETA. They often measure the recipiency rate as the ratio of regular UI claimants (insured
unemployment or IU which includes some not receiving benefits) to total unemployment
(or TU). The IUTU ratio (weekly UI claimants as a proportion of weekly unemployment)
averaged 0.367 between 1967 and 2007 as opposed to the 0.316 for the WBTU ratio
(weekly UI beneficiaries as a proportion of weekly unemployment).
The Role of UI as An Automatic Page 19 July 2010 Stabilizer During a Recession
Chart 2.1 shows the national recipiency rate for the period 1967 to 2008. The chart has
two series: The annual WBTU ratio and the centered five-year average of the WBTU
ratio. The latter series extends only to 2006, the latest available centered five-year
average.
Chart 2.1. Regular UI Recipiency Rates, 1967 to 2008
Year-to-year changes in the recipiency rate11
for the regular UI program can be large, as
clearly shown in the annual series in Chart 2.1. The two series, particularly the five-year
averages, also show a decrease in recipiency during the early 1980s and an increase in the
mid-1990s. In the most recent years, the recipiency rate has returned to levels that
approach the levels of the 1970s.
Within a given year, UI benefit recipiency rates exhibit wide variation across states.
State-level averages of the WBTU ratio during 1967-2007 were below 0.20 in five states
but exceeded 0.45 in four states over the same 41 years.12
11 The WBTU ratio at the state level is first available in 1967. In earlier research, the author has developed state-level estimates of TU for all states starting in 1967.
12 Averages below 0.20 were present in Colorado, Florida, South Dakota, Texas and Virginia. Averages above 0.45 were present in Connecticut, Massachusetts, New Jersey and Rhode Island.
0.20
0.25
0.30
0.35
0.40
0.45
1967 1972 1977 1982 1987 1992 1997 2002 2005 2008
WBTU - Annual WBTU - Centered Five Year Avg.
The Role of UI as An Automatic Page 20 July 2010 Stabilizer During a Recession
But state-level WBTU ratios exhibit quite stable relative rankings. Chart 2.2 helps
illustrate this relative stability. Three of the six included states exhibit consistently high
recipiency rates (Massachusetts, New Jersey, and Pennsylvania) while three exhibit
consistently low recipiency rates (Florida, Texas, and Virginia). For both groups, annual
recipiency varies, with the variation larger for those with high recipiency rates; but, not a
single data point moves a state from one group to the other.
Chart 2.2. Regular UI Recipiency Rates in Six States, 1967 to 2008
The contrast in recipiency rates for the two groups of states would seem to have clear
implications for UI program performance in stabilizing the economy. States with high
recipiency can be expected to exert greater stabilizing effects than states with low
recipiency, given that the replacement rates of high recipiency states are not noticeably
lower than in low recipiency states. The size of the differential effect would also be
influenced by the size of offsetting responses caused by experience-rated UI taxes. These
issues are explored in Chapter 4 using the Economy.com state models.
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
1967 1972 1977 1982 1987 1992 1997 2002 2005 2008
Florida Texas Virginia Mass. N.J. Penn.
The Role of UI as An Automatic Page 21 July 2010 Stabilizer During a Recession
We modeled the recipiency rate for regular UI benefits in each state using four
explanatory variables. The unemployment rate for the current year (TUR for total
unemployment rate) is expected to enter with a positive coefficient, while the lagged
TUR is expected to enter with a negative coefficient. The positive effect of the current
TUR reflects the change in the composition of unemployment when unemployment
increases. The proportion who are job losers increases with higher unemployment, and
job losers are the group most likely to collect UI benefits. The negative effect of the
lagged TUR arises from 1) benefit exhaustions, as those with long benefit duration use up
their entitlements and 2) the effects of reduced base period earnings and monetary
eligibility caused by higher lagged unemployment. These current and lagged effects have
been observed for many years.
UI benefit recipiency has also undergone changes during certain periods. Restrictions on
benefit eligibility occurred in the early 1980s and a downward shift in recipiency has
been widely noted.13
The shift is apparent in Chart 2.1. Policies at the state and national
level were responsible for much of this shift. Less noticed has been an increase in
recipiency that dates from the mid-1990s. Two factors provide at least part of the
explanation for this recent increase: the aging of the labor force and increased reliance by
employers on permanent (as opposed to temporary) layoffs during recessions. The two
trend changes are approximated with dummy variables: The first, D1981 equals zero
before 1981 and 1.0 from 1981; the second, D1996 equals zero before 1996 and 1.0 from
1996. As will be seen, both dummies make significant contributions to explained
variation in state-level WBTU ratios.
In each state, a regression was fitted for the 41 years from 1967 to 2007. Table 2.2
summarizes the regression results; each state regression appears in Table A.2 of
Appendix A. The first thing to note about Table 2.2 is the number of low adjusted R2s.
Nineteen fall below 0.40 and just 12 exceed 0.60. In other words, on average, the
regressions explain less than half the variation in the WBTU ratio over the 1967-2007
13 Several papers have documented a downward shift in recipiency in the early 1980s: Blank and Card (1991), Burtless and Saks (1984), Corson and Nicholson (1988) and Vroman (1991).
The Role of UI as An Automatic Page 22 July 2010 Stabilizer During a Recession
period. The rather large size of the standard errors of the estimates is also apparent. The
average of 0.043 indicates that an increase or decrease of 0.043 in the WBTU ratio from
one year to the next would not be statistically significant in the majority of states. The
regular UI recipiency rate is, thus, a noisy statistical series in individual states.
The coefficients in Panel B are simple averages, but three features are noteworthy. First,
the average constant term, 0.336, is similar to the overall average WBTU ratio of 0.314.
Second, the sizes of the coefficients for the TUR and the TUR lagged are nearly identical
and opposite sign. Recipiency increases when unemployment increases, but the negative
pushback from exhaustions and reduced monetary eligibility in the next year is nearly as
large. Thus, there is no long-run effect on recipiency when unemployment rises or falls
but there is a strong short-run response.14
On average, an increase in the unemployment
rate by one percentage point raises the WBTU ratio by slightly more than two percentage
points in the same year, but the ratio falls by about the same amount during the next year.
Third, the average sizes of the two trend shift dummies (D1981 and D1986) are nearly
identical. The average downward shift in 1981 was 2.7 percentage points and the increase
from 1996 was also 2.7 percentage points. Combined, the coefficients indicate that the
recipiency rate after 1996 had returned to a level close to its average prior to 1981.
14 The size of the averages may seem large to some readers. The TUR and the TUR lagged in the regressions were measured as proportions, not as percentages.
The Role of UI as An Automatic Page 23 July 2010 Stabilizer During a Recession
Table 2.2. Summary of Recipiency Rate Regressions, 1967 to 2007
Panel A. Summary Statistics for 51 Programs
Panel B. Sign and Significance of Coefficients
Positive,
Significant Positive,
Not Signif. Negative,
Not Signif. Negative,
Significant Average
Constant 50 1 0 0 0.336 TUR 36 12 2 1 2.066 TUR Lag 0 0 10 39 -2.193 D1981 4 9 19 19 -0.027 D1996 20 16 10 5 0.027
Panel C. Average Recipiency Rates
WBTU
Number of States Below 0.20 5 0.20-0.249 9 0.25-0.299 11 0.30-0.349 9 0.35-0.399 8 0.40-0.449 4 0.45 Plus 5 Average 0.314
Source: WBTU ratios developed from data published by the U.S. Department of Labor.
Adjusted R2 Standard Error
<0.10 4 <=0.030 7 0.10-0.199 3 0.030-0.0399 17 0.20-0.299 4 0.040-0.0499 17 0.30-0.399 8 0.050-0.0599 5 0.40-0.499 11 0.060-0.0699 3 0.50-0.599 9 0.070 Plus 2 0.60-0.699 6 0.70 Plus 6 Average 0.454 Average 0.043
The Role of UI as An Automatic Page 24 July 2010 Stabilizer During a Recession
2.5 Extended UI Benefits
Besides regular UI benefits, unemployed workers in some states and/or time periods are
also eligible for benefits that extend beyond 26 weeks. There is a permanent federal-state
extended benefits program (EB) that may pay up to an additional 13 weeks of benefits (or
even 20 weeks in certain situations) if a state EB trigger is “On.” Additionally, the
payment of temporary federal benefits (TFB) occurs in certain periods because of federal
UI legislation enacted during recessions. The TFB programs are temporary with definite
“sunset” dates. Both EB and TFB programs were activated in 2008 and both expanded
considerably in 2009, a result of both legislation and higher unemployment rates. During
all earlier recessions, EB has been financed 50-50 by the state and the federal
government, while TFB has been fully federally financed. The American Recovery and
Reinvestment Act (ARRA) of February 2009, however, included a provision to have the
federal partner finance all EB payments for claimants who start to collect EB before
ARRA expires.
The EB and TFB programs have been relatively important in many past recessions (recall
Table 1.1). During 1992 and 1993 the TFB program (termed Emergency Unemployment
Compensation or EUC, the same name as the current TFB program) paid amounts equal
to fully half of regular UI benefits. Between 1971 and 1982, EB made substantial
payments during recessionary years. While EB was not important during the recessions
of 1991 and 2001,15
the number of states paying EB in 2009 increased from three during
the first week of January to 36 to 38 between August and November. One-time financial
incentives under ARRA (full federal financing), changes to temporary TUR triggers, plus
increases in unemployment to higher levels than in the 1991 and 2001 recessions explain
the increase in EB payments by the states during 2009 and 2010. EB during 2009 totaled
$6.1 billion, exceeding $1.0 billion for the first time since 1983.
The current EUC program has been the subject of seven federal legislative enactments
(July 2008, November 2008, February 2009 and November 2009, December 2009, and 15 Only nine states activated EB during and after the 1991 recession; just six activated EB during and after the 2001 recession.
The Role of UI as An Automatic Page 25 July 2010 Stabilizer During a Recession
March 2010, and April 2010). For the first 11 months of 2009, provisions under the
federal stimulus legislation paid EUC for either 20 or 33 weeks depending upon each
state’s TUR. Eligibility for 20 weeks applied if the three-month TUR was at least 6.0
percent, and for 33 weeks if the TUR was at least 8.0 percent. States eligible for 33 weeks
have increased from 20 during the first week of January 2009 to 47 during October 2009.
Because of the November 2009 legislation, all states could pay at least 34 weeks of EUC
during the final weeks of 2009. As of May 2010, there are four separate tiers of EUC
with maximum potential EUC duration of 53 weeks in over 30 states.
Because EB was not active in most states during the 1991 and 2001 recessions, recent
information on the relative importance of EB benefits was lacking for most states early in
2009. As noted, however, in late 2009 about three states in four were paying EB. The EB
and EUC provisions of current federal UI legislation will run through early November
2010. If the economic recovery proceeds slowly and the recession extends well into 2010
and later, further EB and/or EUC extensions are possible (even likely). Thus, the
performance of the regular UI program under alternative future scenarios can be
estimated with much greater confidence than the performance of EB and EUC.
Discussion of the simulations of the EB and EUC programs in Chapter 4 are careful in
describing the underlying assumptions regarding when they are “On.”
2.6 Regular UI Replacement Rates
The replacement rates to be used in the simulation analysis are from the Unemployment
Insurance Financial Handbook, i.e., the ratio of the average weekly benefit for full weeks
of unemployment to the average weekly wage of taxable plus reimbursable employers.
Since 1967, this ratio has varied between 0.329 and 0.377 at the national level.
In contrast to the recipiency rate, the multiple regressions are quite successful in
explaining the replacement rate. Table 2.3 summarizes state-level regressions that span
the years 1967 to 2007. (The individual state regressions appear in Table A.3 of
Appendix A.) Among the 51 state-level regressions in Table 2.3, 38 have adjusted R2s of
The Role of UI as An Automatic Page 26 July 2010 Stabilizer During a Recession
0.70 or higher, while just four explain less than half the time series variation in the
replacement rate. Also indicative of generally good explanatory power, the regressions
usually have small standard errors. More than half (27) are smaller than 0.012, while just
11 exceed 0.016. The average standard error of 0.0126 is less than one-third the average
for the recipiency rate regressions summarized in Table 2.2 above.
For individual states, several factors make significant contributions to explaining
replacement rate variation. Nearly all regressions include three explanatory variables: 1)
the ratio of the maximum weekly benefit to the average weekly wage (MxBenAWW), 2)
the TUR, and 3) the TUR lagged. Note that all 51 MxBenAWW variables enter with a
positive and significant coefficient. This variable was the most important contributor to
explained variation in 45 of 51 regressions. When the maximum weekly benefit increases
relative to average wages, the replacement rate increases. The current unemployment rate
(TUR) exhibits a uniformly positive coefficient in Table 2.3, which is significant in 37
states. In contrast, the lagged TUR enters negatively with a significant coefficient in 31 of
43 states. This variable was not used in eight states because of collinearity with the
current TUR. When both were entered in these states, neither was significant and there
was no improvement in the overall fit, i.e., the adjusted R2.
Three other influences on the replacement rate entered significantly in a number of states.
The statutory replacement rate changed in 15 states during the 1967-2007 period. All 15
slopes had the expected positive signs, of which all but one were significant. Most states
operated with a single statutory replacement rate during these years.
Most states determine a claimant’s weekly benefit using high quarter earnings from the
base period. Over the 1967-2007 period, however, several states changed their WBA
calculation from using the single high quarter of earnings in the base period to using
average earnings from the two highest quarters. In nearly all instances, this change
reduced the weekly benefit and the associated replacement rate. Note in Panel B that
seven of the eight coefficients for the two-quarter calculation (D 2Qtr) are negative and
six are significant. On average, the move to a two-quarter calculation reduced
The Role of UI as An Automatic Page 27 July 2010 Stabilizer During a Recession
replacement rates by 0.02. A second change that reduced replacement rates was the
change to an average weekly wage calculation from a high quarter calculation (or vice
versa). The associated dummy variable (D AnnWage) was set at 1.0 in years when the
annual wage calculation was used and 0.0 when the high quarter calculation was used. In
eight of 10 states, this dummy coefficient had the expected negative sign, of which five
were significant. The two exceptions were New York and Wisconsin. Both states
changed to a high quarter calculation, but the replacement rate in both was lower in the
post-change period. No good explanation for this result has been found. Discussions with
professional staff in the two states did not help in finding a solution.
The Role of UI as An Automatic Page 28 July 2010 Stabilizer During a Recession
Table 2.3. Summary of Replacement Rate Regressions, 1967 to 2007
Panel A. Summary Statistics for 51 Programs
Adjusted R2 Standard Error
<0.50 4 0.006-0.0099 14 0.50-0.599 3 0.010-0.0119 13 0.60-0.699 6 0.012-0.0139 9 0.70-0.799 12 0.014-0.0159 4 0.80-0.899 17 0.016-0.0179 5 0.90 Plus 9 0.018 Plus 6 Average 0.772 Average 0.0126
Panel B. Sign and Significance of Coefficients
Positive, Significant
Positive, Not Signif.
Negative, Not Signif.
Negative, Significant Number Average
Constant 34 4 8 5 51 0.080 MxBenAWW 51 0 0 0 51 0.439 TUR 37 14 0 0 51 0.623 TUR Lag 0 0 12 31 43 -0.488 RRate Stat 14 1 0 0 15 0.486 D 2Qtr 0 1 1 6 8 -0.020 D AnnWage 2 0 3 5 10 -0.019
Panel C. Average Replacement Rates and Maximum Benefit to AWW Ratios
Repl. Rate 1967-07 1998-07 MxBenAWW 1967-07 1967-97 1998-07
Below 0.33 9 13 Below 0.35 2 2 3 0.33-0.349 8 4 0.35-0.399 6 5 6 0.35-0.369 13 8 0.40-0.449 12 9 11 0.37-0.389 6 9 0.45-0.499 13 18 9 0.39-0.409 8 8 0.50-0.549 8 10 7 0.41-0.429 4 4 0.55-0.599 9 6 7 0.43 Plus 3 5 0.60 Plus 1 1 8 Average 0.366 0.364 Average 0.474 0.469 0.488 Source: Handbook replacement rates published by U.S. Department of Labor. Other variables derived by the author from data published by the Office of Workforce Security and BLS.
The Role of UI as An Automatic Page 29 July 2010 Stabilizer During a Recession
The bottom panel in Table 2.3 summarizes the distribution of replacement rates and the
ratio of the maximum weekly benefit to the average weekly wage, with attention to the
last 10 years (1998-2007) as well as the full 1967-2007 period. Note that the average
replacement rate was essentially the same in the last decade as for the full period. The
MxBenAWW ratio did increase somewhat in the most recent period, but the increase in
the 51-state average was only 4.1 percent compared to the 1967-1997 period. The
regressions of Table 2.3 and the back-up detail of Appendix Table A.3 suggest that the
determinants of replacement rates are known and that no important trends were present
during the 41-year sample period examined here.
The summary provided in Panel C of Table 2.3 also points to a shortcut that can be used
in the simulation analysis. Since the replacement rates exhibit comparatively small
variation, the simulations can legitimately use average state-level replacement rates as an
alternative to the regression equations displayed in Table A.3. The simulation results to
be summarized in Chapter 4 take this simpler approach, using as state-level replacement
rates a 10-year average.
2.7 Summary
This chapter examined behavioral relationships that are central to understanding the
performance of UI programs in individual states. Multiple regressions were used to
characterize time series variation in average UI tax rates and in recipiency rates and
replacement rates in the regular UI program. (The state-level regression results are
displayed in Appendix A.) The chapter also described a cross-section analysis of
differences in UI tax rates across 19 major industries in each state. All these relationships
have been entered into the Economy.com state-level simulation models. Chapter 3
describes the Economy.com state models that underlie the simulation results to be
presented and discussed in Chapter 4.
The Role of UI as An Automatic Page 30 July 2010 Stabilizer During a Recession
CHAPTER 3. MODELING THE MACROECONOMIC EFFECTS OF
UNEMPLOYMENT INSURANCE
Our analysis of UI as an automatic stabilizer was conducted using the macroeconomic
models developed by Economy.com, a branch of Moody’s Investor Services
Incorporated. This chapter describes the structure of those models and discusses the
strategy followed in the simulation analysis.
3.1 The Economy.com Model
Economy.com has developed econometric models suitable for analysis at the national,
state, and MSA levels of geographic detail. Our simulations used state models for all 50
states plus the District of Columbia (hereafter 51 states). This geographic detail matches
the UI program’s structure, with benefit and financing provisions set by the states and
differing noticeably from state to state.
Economy.com models use quarterly seasonally adjusted data with quarterly flows
measured at annual rates. They carry historic values back at least 20 years and can make
forecasts for as many as 30 future years. In our analysis, many simulations were extended
to 2020, or 12 years beyond 2008, the most recent year with fully available annual data.
This capacity to make lengthy future projections is important because the UI tax rate
relationships have four-year lags on benefit payments. Thus, recession-related increases
in benefits of 2009, 2010, and later years will affect UI taxes through 2014 and beyond.
The models easily incorporate these lagged effects.
3.2 Model Structure
The state models characterize each state economy in six areas: 1) demographics, 2) labor
market-real gross product, 3) personal income and average earnings, 4) credit and
banking, 5) real estate and housing and 6) consumer demand. Several state-specific
relationships are included in each of these areas (or modules), as described in a paper by
The Role of UI as An Automatic Page 31 July 2010 Stabilizer During a Recession
Cochrane (2006). The following paragraphs give a brief summary of structural features
and key relationships.
Each state model has a complete demographic sector that updates state population
estimates with projections of migration, births, and deaths. The total population is divided
into age cohorts, and population change includes certain age-specific relationships. Net
migration is determined by recent rates of job creation and the change in state
unemployment relative to the national average. Separate relationships determine in-
migration and out-migration. If aggregate state economic performance is below average,
both these population flows respond and slow the pace of statewide population growth.
International and domestic population flows are incorporated into the state models.
Paralleling the model’s population dynamics are changes in the number of households.
Households are disaggregated by age of head and changes are linked to state population
growth. Labor market conditions also influence the total number of households. Higher
unemployment reduces the rate of new household formation.
Central to each state model is the determination of real output (Gross State Product or
GSP). Estimates of GSP are available from the Commerce Department’s Bureau of
Economic Analysis (BEA) by detailed NAICS16
industries. State-level GSP for each
industrial sector is linked to national GDP in that industry, with adjustments made
according to each industry’s cost of doing business. This cost variable is discussed below
and in Appendix B. State-level GSP for industries in the service sector is driven primarily
by local demand conditions, where the size of the state’s population and the level of
personal disposable income are two key determining factors. Establishment employment
is linked to real output through derived demand relationships.
Personal disposable income has wages and salaries as its largest component, but it
includes all the other components from the national income accounts as well.
Specifically, personal disposable income includes dividends, interest, rents, proprietors’
16 NAICS – North American Industrial Classification System
The Role of UI as An Automatic Page 32 July 2010 Stabilizer During a Recession
income, and Government transfer payments to persons less personal taxes. For the
present analysis, transfer payments explicitly recognize each of the three tiers of UI
benefit payments as well as the aggregate of all other transfer payments. While this report
emphasizes the stabilizing effects of UI benefits, it is important to remember that UI
benefits are a small component of total transfers; all other transfers have represented
about 98 percent of total transfer payments in recent years.
Real output is also affected by the cost of doing business (CDB) in each state-industry
sector. The Economy.com state models recognize three areas of costs that contribute to
the overall cost profile for each state-industry sector: Labor costs, energy costs, and tax
burden.17
Labor costs are measured as total wages and salaries (payroll) from the
National Income and Product Accounts (NIPA). To recognize labor productivity growth,
NIPA payroll is deflated by real GSP. Energy costs are estimated as an average of
commercial and industrial electricity prices measured in cents per kilowatt-hour (each
normalized by their respective national average) and the weights provided by national
expenditures for the two types of energy. The calculation of tax burden incorporates
personal, property, and corporate taxes. Taxes also include employer payroll-based
contributions for UI and workers’ compensation. This comprehensive measure of
business plus personal taxes is expressed as a ratio to personal income in the state. Each
state-level tax burden ratio is then measured relative to the national ratio.
The aggregate CDB cost measure is then derived as a weighted average of its three
constituent components. The national weights are 0.75 for labor costs, 0.15 for energy
costs, and 0.10 for tax burden. The weights vary by industry and state. States with an
above-average CDB will experience a drag on real GSP growth over the long run,
particularly in the industrial sectors, as location decisions respond to cost differentials.
The employer taxes that support the UI program enter the Economy.com models through
the CDB cost variable. States with above-average UI taxes and an associated high CDB
17 See Appendix B for a fuller description of how the cost of doing business is measured. Essentially, it is a weighted average of costs by major cost categories.
The Role of UI as An Automatic Page 33 July 2010 Stabilizer During a Recession
will experience some loss of real output due to costs. States with high unemployment
rates and/or high UI benefit payments per unemployed worker will be subject to this
negative effect on real output.
Real demand and output in each detailed industry and industry productivity are the main
determinants of employment in each industry. The models have separate regression
relationships that link employment to real output through a derived demand for labor
relationship. Through this mechanism, the financing of the UI program has negative
output and employment consequences for a state.
UI benefit payments, in contrast, have a positive effect on real output and employment.
These transfer payments increase household disposable income and consumption.
Increases in UI benefits have an immediate effect upon disposable income and
consumption expenditures. Unlike higher UI taxes, which operate with a long (four-year)
lag, increases in benefit payments (from all three tiers of UI benefits) immediately raise
household income. These transfer payments are then almost entirely spent on
consumption items in the same year.
Thus, the two channels whereby UI affects the rest of the economy are through increases
in consumption from UI benefits and increases in UI taxes (which reduce real output in
affected industries by raising employer costs).
3.3 The Simulation Strategy
To examine the effects of UI on the macro economy, several different determinants (or
treatments) were included in our analysis. Four separate elements of the UI program can
influence the time path of real GDP, total employment, and total unemployment. These
four are: (1) regular UI benefits, (2) temporary (or emergency) federal benefits (EUC in
2008, 2009, 2010, and perhaps later), (3) Federal-State Extended Benefits (EB), and (4)
UI taxes. For the present report, the EUC and EB programs are modeled as a single
extended benefit program. Even with the modifications of the EB triggers made under the
The Role of UI as An Automatic Page 34 July 2010 Stabilizer During a Recession
fiscal stimulus package of February 2009, the bulk of all extended benefits in 2009 were
EUC benefits. The EB component of their combined total for 2009 was only about 10
percent.
Thus, the analysis examines the effects of three separate components of the UI program:
regular UI benefits, extended benefits (EUC plus EB) and UI taxes. It should be noted
that the UI taxes included in the analysis are the state taxes that support payment of
regular UI benefits. The federal taxes that support program administration are not
modeled. This approach also assumes that the full costs of EUC and EB payments are
supported by the federal partner and add to the federal budget deficit. The effects of EUC
and EB on the deficit are measurable in 2009, adding some $40 billion to the deficit.
Their effects in financial markets are the same as other categories of deficit-increasing
expenditures. No explicit treatment of the feedback effects of the deficit on macro
performance is included in this analysis.
From the perspective of the business cycle, the UI program is important in stabilizing the
time paths of macro variables like real GDP and total employment. To gauge UI’s
stabilizing impact we simulate a steady growth counterfactual and examine downward
deviations from the counterfactual. The counterfactual projects macro variables under an
assumption of reasonably steady growth during and after the periods affected by the
recession, which officially began in the fourth calendar quarter of 2007. While the current
recession may officially end in 2010, it is clear that unemployment will remain high and
real output will remain considerably below potential real output for several future years.
Associated with high unemployment will be elevated levels of UI benefit payments and
UI taxes.
The growth counterfactual to be used is a growth projection from the national
Economy.com model used in the fourth calendar quarter of 2007. This foresees annual
real GDP growth in later periods of between 2.7 and 4.7 percent, with growth in most
years above 3.0 percent. During the years between 2008 and 2020, the unemployment
rate is projected to range between 4.2 and 5.1 percent (lower in the later years) and
The Role of UI as An Automatic Page 35 July 2010 Stabilizer During a Recession
average 4.5 percent. This path approximated full employment growth as projected by the
2007Q4 Economy.com model.
The steady growth path is then compared to a time path that approximates a deep
recession. Rather than developing an artificial recessionary time path, we used the
historic time path of the economy (2007Q4 through 2009Q2) for comparison with the
steady growth path. For quarters starting in 2009Q3, the macro time path follows what
Economy.com projects as the most likely future time path for the economy.18
Our model-based analysis derives estimates holding constant many other factors in the
economy. The research strategy is to focus on the three aspects of the UI program
(regular UI benefits, extended benefits and EUC, and UI taxes) in both the steady growth
environment and in the recessionary environment. To do this, we simulate the effects of
each factor in such a way that its separate contribution to macro performance can be
isolated. Thus, the effects of regular UI benefits are simulated first under the assumption
of no EB or EUC program, and benefit payouts are simulated with and without UI taxes.
The extended benefit programs are then added to regular UI to yield estimates of their
marginal effect in addition to that of regular UI. Because UI taxes operate with long lags,
these are then added to the simulations to produce results with all aspects of the UI
program activated.
The method of holding constant the effects of variables not included in a particular
simulation is to keep that variable constant in real terms throughout all future periods.
Where it is appropriate in Chapter 4, we discuss further the details of how variables are
treated in specific simulations.
18 To avoid using confusing terminology, this time path will be termed the “future time path”. When Economy.com provides forecasts for its clients, it projects five different future time paths, three of which are more pessimistic than the time path it deems most likely. The most likely time path is judged to have a 50 percent probability of occurring. For its clients, this most likely time path is termed the “baseline time path”, but we will not use this terminology in the present report.
The Role of UI as An Automatic Page 36 July 2010 Stabilizer During a Recession
CHAPTER 4. SIMULATION RESULTS
This chapter summarizes the results of simulations of the U.S. economy with and without
UI during a severe recession as experienced in recent years. Within the deep recession
simulations, four sets of results are summarized.
• Path 1 traces the time paths of macro variables when regular UI benefits and
associated taxes respond to changes in unemployment.
• Path 2 adds benefit payments from two extended benefit programs: Emergency Unemployment Compensation (EUC) and Federal-State Extended Benefits (EB).
• Path 3 traces the time path of macro variables with just regular UI benefits
responding to unemployment (UI taxes not responding).
• Path 4 traces the time path when both regular UI benefits and regular UI taxes are held constant in real terms at their pre-recession level.
By comparing Path 2 with Path 1, one can assess the additional stabilizing effects of
extended benefits. By comparing Path 1 with Path 3, one can estimate the extent to which
UI taxes lessen the stabilizing effect of regular UI benefits. By comparing Path 4 and
Path 3, one can assess the response of regular UI benefits to the recession and how much
the time paths of real GDP and employment differ when regular UI benefits respond.
The four recession time paths are simulated for 51 “state” programs (i.e., including the
District of Columbia).19
19 Economy.com does not support complete models for Puerto Rico and the Virgin Islands, two other jurisdictions within the state UI program.
The simulations extend through 2020, but primary emphasis is
placed on results that extend through 2010Q2. By the end of 2010, real output and
employment have started to increase while the unemployment rate has also stabilized and
started to decline. Given the amount of state-level and time period detail generated for
each variable, the chapter necessarily summarizes the results at a high level of
aggregation. Again, the model uses quarterly data for GDP, UI benefits and UI taxes but
all measured at annual rates.
The Role of UI as An Automatic Page 37 July 2010 Stabilizer During a Recession
After examining the stabilizing role of the regular UI program, the model then adds the
extended benefit programs to estimate the added stabilizing effects that they provide.
While separate detail for EUC and EB is generated in these “extended benefit”
simulations, the text emphasizes just the combined effects of EUC and EB. Readers
should understand, however, that the vast majority of extended benefits are EUC benefits.
In historic data currently available (through the early months of 2010), the highest
quarterly payout of EB was the $2.8 billion paid during the third quarter of 2009 while
EUC benefits totaled $11.4 billion during the same quarter. Between 2008Q3 and
2009Q4 cumulative EUC benefits totaled $51.0 billion while cumulative EB benefits
totaled $6.9 billion, or about 12 percent of their combined total.
While the recession simulations are of principal interest, the steady growth simulations
provide one way to gauge the impact of a recession on real output and employment. The
steady growth simulations are described first.
4.1 The Steady Growth (No-recession) Simulations
As noted, the performance of the UI program should be measured against a
counterfactual simulation where the economy experiences steady growth. This
simulation has three variants: 1) growth with regular UI benefits and taxes functioning, 2)
growth with regular UI benefits responding but UI taxes constant, and 3) growth with
regular UI benefits and taxes both constant in real terms. Each of these three simulated
paths has a counterpart in later simulations where the recession started in 2007Q4 and
follows a recessionary path into later periods. The recession path closely approximates
the economy’s actual path through the second quarter of 2009 and then follows the most
likely path (as judged by Economy.com) for later quarters through 2020.
The three time paths of the baseline scenario are depicted in Chart 4.1. Note that their
proximity is practically identical. The lines are so similar that the graph does not display
three distinct series. In a situation where unemployment varies within a narrow range, the
The Role of UI as An Automatic Page 38 July 2010 Stabilizer During a Recession
quantitative effect of the benefits and taxes of the regular UI program are very small. The
chart shows that with just benefits but no UI taxes, real output is on the highest path as
would be expected, but the differences are tiny. Real UI benefits average 0.23 percent of
real GDP between 2007Q4 and 2020Q4.
Chart 4.1 Steady Growth, Real GDP Time Paths, 2008 to 2020
Source: Simulations with the Economy.com model. Data in billions of 2000 dollars.
While the aggregate real GDP growth paths are very similar, clear differences in the size
of benefit payouts are observed in state-level data. In comparisons to be repeated later in
the chapter, the ratios of real benefits to real output across the states revealed large
contrasts. For the 10 states with the highest recipiency rates, real regular UI benefits
averaged 0.38 percent of real GDP compared to 0.12 percent for the 10 with the lowest
recipiency rates. These contrasts are sizeable even in a steady growth scenario. When the
source of the contrast is examined, it is found to be differences in recipiency rates (the
ratio of UI beneficiaries to unemployment). The unemployed in high-recipiency states are
more than twice as likely to receive regular UI benefits when compared to the
unemployed in low-recipiency states. The simple averages of the recipiency rates for the
10,000
12,000
14,000
16,000
18,000
20,000
Benefits and Taxes Benefits Only Real Benefits Constant
The Role of UI as An Automatic Page 39 July 2010 Stabilizer During a Recession
two groups of states in 2007 were 0.470 versus 0.193. In contrast, there is very little
difference in the average replacement rates (weekly benefits divided by weekly wages).
During 2007, the average replacement rate across the 10 high-recipiency states was 0.348
while it was 0.338 across the 10 low-recipiency states.20
The preceding comparison of states with highest recipiency versus those with lowest
recipiency provides a convenient way to summarize state-level detail without explicitly
displaying 51 state statistics for a particular variable such as real GDP. Recall from Chart
2.2 of Chapter 2 that multiyear patterns of recipiency rates are quite stable for individual
states. Focusing upon states at the extremes of the recipiency rate distribution provides a
convenient way to highlight contrasts among the state programs. This device for
summarizing contrasts across the states will be employed later in the chapter.
4.2 Recession Simulations
In 2008-2009, the U.S. economy experienced the most serious recession of the post-
World War II years. Many observers are describing this period as the “great recession”.
The national unemployment rate averaged 10.0 percent during October-December 2009
and 16 state-level unemployment rates exceeded 10.0 percent during the same quarter.
The annual average U.S. unemployment rate (TUR) for 2009 was roughly twice its level
in 2007 (i.e., 9.3 percent compared to 4.6 percent).
While recession-related increases in unemployment have occurred in all states, the most
severe increases in unemployment and associated reductions in employment have
occurred in the states from two of the nine U.S. Census Bureau divisions, the East North
Central and Pacific divisions. Monthly unemployment rates during 2009 have averaged at
least a full percentage point and often two percentage points above the national average.
The national average unemployment rates for the two divisions were respectively 10.7
percent and 11.1 percent. Higher unemployment has most severely affected youths,
minorities, men, and those with low educational attainment. 20 High recipiency states: Alaska, Connecticut, Idaho, Massachusetts, New Jersey, Pennsylvania, Rhode Island, Vermont, Washington D.C., and Wisconsin. Low recipiency states: Arizona, Colorado, Florida, Mississippi, New Hampshire, Oklahoma, South Dakota, Texas, Utah, and Virginia.
The Role of UI as An Automatic Page 40 July 2010 Stabilizer During a Recession
Chart 4.2 displays six time series (three pairs) of unemployment rates (TURs) from the
Economy.com state model. These quarterly data cover the seven years 2007 to 2013 and
all series are from simulations performed for this research project. The national series are
actual historic data from 2007Q1 to 2009Q2, while later periods are model-based
projections. The chart displays averages for the 10 states with highest recipiency and the
10 with lowest recipiency. The time profiles of the three steady-growth (or no recession)
series and the three recession series are very similar. In the recession series, the peak
unemployment rate is reached in 2010Q2. This quarter also has the highest
unemployment rate for 39 of the 51 state-level projections.21
Chart 4.2. Unemployment Rates in the Recession and No Recession Scenarios, 2007Q1 to 2013Q4
Source: Simulations with the Economy.com state model. Unemployment as a percent of the labor force.
21 Of the 12 states where the highest TUR occurs in another quarter, that quarter is 2009Q4 for three, 2010Q1 for four, and 2010Q3 for four. The other highest TUR occurs in Texas in 2011Q2.
The Role of UI as An Automatic Page 41 July 2010 Stabilizer During a Recession
Note in Chart 4.2 that both 10-state average TURs are below the national average,
particularly in the low-recipiency states. It should also be noted that the TURs in the
recession simulations remain above those in the no recession simulations not just through
2013 but also for all years through 2020 (not shown).
The sizes of the real output and employment losses are noteworthy. In 2010Q2, the
quarter of peak unemployment, real GDP in the recession simulation is 7.5 percent lower
than in the no-recession simulation. Total employment during 2010Q2 in this simulation
falls 9.6 million (6.7 percent) below employment in the no recession simulation and the
TUR is more than double its no-recession counterpart (10.38 versus 4.62 percent). These
large declines in real GDP and employment help to point out the need for having a robust
UI program to offset the recession’s negative effects on families and individuals.
While the decrease in real output during 2008-2011 is to be expected, the recession also
lowers real GDP in all later periods of the 2010-2020 decade. This impact on the growth
path arises in part from reduced business fixed investment during the recession, which
reduces the size of the capital stock. In the Economy.com model, the recession has long-
run effects on real GDP and employment as well as short-run effects.
It may be instructive to describe the size of the reductions in real GDP caused by the
recession. In 2008Q2, the downward deviation from the steady growth path projected in
2007Q4 is 0.9 percent, but it then rapidly increases to 3.5 percent at the end of 2008, 6.5
percent at the end of 2009, and 8.0 percent at the end of 2010. The downward deviation
then decreases, but only to 7.8 percent at the end of 2011 and 6.7 percent at the end of
2012. After 2012, the convergence of the recession time path towards the no recession
time path ceases. The effect of the recession on the growth path, in other words, is very
large. The deviation in real GDP between the no-recession and the recession growth paths
during the three recession-impacted years 2009-2011 averages $905.5 billion. This
represents about 7.0 percent of no-recession real GDP.
The Role of UI as An Automatic Page 42 July 2010 Stabilizer During a Recession
The contrasting growth paths are strongly influenced by four changes made in the
Economy.com model between the 2007Q4 version and the 2009Q2 version. In light of
economic developments during late 2008, the changes identified below were made to the
forecasting model that had been used at the end 2007. In later discussions, the “no-
recession model” used at the end of 2007 will be termed the “2007Q4 version”, and the
“recession model” used in mid-2009 will be termed the “2009Q2 version”. Key
differences between the two models are the following.
1) The future growth rate of the labor force was reduced.22
2) The full employment unemployment rate was revised upward from 4.2 percent of the labor force to 5.5 percent.
23
3) Household savings rate was revised upward to 7.5 percent of household disposable income, an increase from 6.5 percent.
4) The recession reduced business fixed investment, hence the size of the total stock of machinery and equipment.
All four factors combine to produce lower growth paths for real GDP and employment
and a higher unemployment rate during the 2010-2020 decade. As a result, the post-
recession growth path of the 2009Q2 model remains substantially below the no-recession
steady growth path of the 2007Q4 model.
4.3 Regular UI Benefits
The recession causes a large response in UI benefit payments. Regular UI benefits
increase noticeably in 2008Q1 and grow strongly over the next six quarters. Nominal
benefits (measured at an annual rate) increase from $39 billion in 2008Q1 to $96 billion
in 2009Q3 and 2009Q4. Thereafter regular UI benefits decrease as the economy recovers
and unemployment moves downward. Total nominal benefits in 2010Q4 decrease to $76
billion and then to $56 and $45 billion at the end of the following two years. Measured in
real terms (deflated by the GDP deflator based on the year 2000) real benefits in 2012Q4
are $35.1 billion, roughly the same as during 2008Q2. Because the unemployment rate 22 Labor force growth in the 2010-2020 decade was reduced substantially, from 1.3 percent per year to 0.7 percent. The analysis of changing labor force participation patterns is summarized in Marisa Di Natale and Sophia Koropeckyj, “Forecasting U.S. Labor Force Participation,” Moody’s Regional Financial Review, (November 2007), pp 20-27. 23 The change reflects an assumed increased rate of worker dislocation from jobs and reduced geographic mobility due to the decline in the value of homes and an associated reluctance to move.
The Role of UI as An Automatic Page 43 July 2010 Stabilizer During a Recession
never returns to its pre-recession level, real regular UI benefit payments never fall below
$33 billion.
Chart 4.3 displays the quarterly time paths of real regular UI benefits and UI taxes
(nominal values deflated by the GDP deflator) from the start of the recession (2007Q4) to
the period when unemployment stabilizes at 5.4 percent (2013Q4). Note how real
benefits increase sharply during the first three calendar quarters of 2009 and then descend
gradually after 2009Q4. By the end of 2013, real regular UI benefits have returned to
their level of early 2008. Chart 4.3 also clearly displays the response of UI taxes.
Aggregate real UI taxes start to exceed $30 billion in 2009Q4, reach a peak above $60
billion in 2012Q2, and then start decreasing. This time pattern reflects the long (4-year)
lags in the tax rate equations discussed in Chapter 2.
Chart 4.3. Real Regular UI Benefits and UI taxes, 2007Q4 to 2013Q4
Source: Simulations with the Economy.com state model. Data in billions of 2000 dollars.
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
Real Regular UI Benefits Real UI Taxes
The Role of UI as An Automatic Page 44 July 2010 Stabilizer During a Recession
The recession simulation with regular UI benefits responding to unemployment shows a
strong response in all states. Comparing the 10 states with the highest with the 10 with
the lowest recipiency rates, the percentage response of benefits is larger in states with low
recipiency. Between 2007Q3 (the pre-recession quarter) and 2010Q2 (the period of
highest unemployment) real benefits grew by 136 percent nationally (from $28.1 to $66.3
billion at an annual rate). Over the same period, the respective growth percentages for the
high-recipiency and the low-recipiency states were 113 and 232 percent.
Contributing to the increase in regular UI benefit payouts in 2008 and 2009 is a
measurable increase in the recipiency rate as unemployment increases. Nationally the
recipiency rate increases from 0.32 in 2007Q3 to 0.39 in 2009Q3 before starting to
decline. By 2010Q2, the recipiency rate has declined to 0.32, its pre-recession level. For
the 10 high-recipiency states the increase in the recipiency rate between 2007Q3 and
2009Q3 is much smaller (from 0.47 to 0.50), and the subsequent decrease to 2010Q2 is
larger (from 0.50 to 0.42). For the 10 low-recipiency rate states, the average recipiency
rate in 2007Q3 is 0.19, growing to 0.28 by 2009Q3, and then decreasing to 0.24 in
2010Q2. On average, the negative feedback from lagged unemployment onto recipiency
in the current year is stronger in high-recipiency states when compared to the low-
recipiency states. As a result, the recipiency rate decreases more in the later periods of a
recession in high-recipiency states when compared to low-recipiency states.
Following the onset of a recession and the associated increase in benefit payouts, a
negative feedback occurs in regular UI benefit payouts due to benefit exhaustions. Since
maximum potential benefit duration is 26 weeks in all but two states,24 this negative
feedback starts to affect recipiency even before the highest unemployment rate is
reached. The simulations provide strong evidence of this negative feedback.25
24 In Massachusetts and Montana, the maximum durations are 30 and 28 weeks respectively.
Nationally
the highest volume of real regular UI benefit payouts occurs during 2009Q4, as it does
for both the 10 high-recipiency and the 10 low-recipiency states. By 2010Q2, real regular
UI payouts nationwide had decreased by 14 percent from their peak in 2009Q4 (or by
25 This negative feedback is present in nearly every state. See Panel B in Table 2.2 of Chapter 2 and Table A2 in Appendix A which displays the recipiency rate regression equations for each state.
The Role of UI as An Automatic Page 45 July 2010 Stabilizer During a Recession
$10.9 billion at an annual rate). The comparable decreases in the 10 high-recipiency and
the 10 low-recipiency states were 13 and 8 percent, respectively.
The presence of regular UI benefit payments measurably reduces the severity of the
economic downturn. To estimate the size of this effect, we compare two time paths of
real GDP:
1) A recession where regular UI benefits respond to the decrease in real GDP and the increase in unemployment.
2) A recession where regular UI benefits are held constant in real terms.
Note that the second time path allows UI benefits to increase, but only in line with
changes in the GDP price deflator. Also, rather than remove all regular UI benefits and
cause a large negative effect on aggregate demand, output, and employment, this
procedure allows the component of demand coming from the volume of pre-recession UI
benefit payments to be unchanged during the recession.
Chart 4.4 displays the three time paths for the period 2007Q4 to 2010Q2. The failure of
real output to return to the no recession time path projected by the Economy.com model
of 2007Q4 was discussed previously. This raises a question of how to project the level of
real GDP in a no-recession environment. The growth parameters in the 2007Q4 model
were more optimistic than in the recession model of 2009Q2. The simple expedient that
underlies Chart 4.4 is to assume a quarterly growth rate of 0.6 percent (2.4 percent annual
growth) and project this growth for every quarter starting in 2007Q4. This top line gives
the reader a guide to the size of the decline in output. Because the no-recession time path
is not a model-based projection, the shortfall of actual GDP from the no-recession GDP
should be viewed as illustrative of the recession-related decline in real output. The
downward deviation is large, averaging about $800 billion in 2009 and 2010. An
indication of the seriousness of the recession is that real GDP does not return to its level
of 2007Q4 until the middle of 2010.
The Role of UI as An Automatic Page 46 July 2010 Stabilizer During a Recession
Chart 4.4. Three Time Paths of Real GDP, 2007Q4 to 2010Q2
Source: Recession time paths from the Economy.com model. The no-recession time path derived at the Urban Institute assuming 0.6 percent growth per calendar quarter. Data in billions of 2000 dollars.
Chart 4.4 shows that regular UI benefits have a stabilizing effect, with real GDP
consistently higher when real UI benefits respond rather than remaining constant.
Between 2008Q3 and 2010Q2, for example, real GDP averages $71 billion higher when
real benefits respond compared to constant real benefits.
Table 4.1 summarizes the time paths of real GDP and real regular UI benefits from 2007
to mid-2010. Columns [1]-[3] display quarterly data on the three time paths of real GDP
shown in Chart 4.4. Column [4] shows the deviation between the stable growth scenario
and the time path where real regular UI benefits do not respond to the recession. This
deviation averages $800 billion during 2009 and 2010.
10,800
11,000
11,200
11,400
11,600
11,800
12,000
12,200
12,400
No Recession Regular UI Benefits Respond Regular UI Benefits Constant
The Role of UI as An Automatic Page 47 July 2010 Stabilizer During a Recession
Table 4.1. Time Paths of Real GDP and Real Regular UI Benefits, 2007Q1 to 2010Q2
Real GDP, Stable
Growth [1]
Real GDP, Recession, Regular UI Responds
[2]
Real GDP, Recession, Regular UI Constant
[3]
Stable Growth
Less Const. UI
[1]-[3] [4]
Real GDP, Responsive
UI less Const. UI
[2]-[3] [5]
Real Regular Benefits
[6]
Change in Benefits
From 2007Q3
[7]
Real GDP Deviation/ Real Ben. Deviation
[5]/[7] [8]
Share of Deviation
Reduced by Regular UI
[5]/[4] [9]
2007Q1 11,424 11,424 11,424 0 0 27.0 - 2007Q2 11,370 11,370 11,370 0 0 27.5 - 2007Q3 11,434 11,434 11,434 0 0 28.1 - 2007Q4 11,503 11,432 11,425 78 7 29.6 1.5 4.5 0.09 2008Q1 11,572 11,453 11,441 131 12 32.0 3.9 3.2 0.09 2008Q2 11,641 11,557 11,537 104 20 35.5 7.4 2.6 0.19 2008Q3 11,711 11,553 11,526 185 27 39.4 11.3 2.4 0.15 2008Q4 11,782 11,414 11,377 404 37 45.1 17.1 2.2 0.09 2009Q1 11,852 11,285 11,226 626 59 61.6 33.5 1.7 0.09 2009Q2 11,923 11,222 11,148 775 74 71.7 43.6 1.7 0.10 2009Q3 11,995 11,289 11,203 792 86 76.7 48.7 1.8 0.11 2009Q4 12,067 11,300 11,207 860 93 77.2 49.2 1.9 0.11 2010Q1 12,139 11,361 11,267 873 94 70.2 42.2 2.2 0.11 2010Q2 12,212 11,434 11,338 874 95 66.3 38.3 2.5 0.11
2008Q3 -
2010Q2 Av. 11,960 11,357 11,287 674 71 63.5 35.5 2.0 0.11
Source: Simulations with the Economy.com model. Data measured in billions of 2000 dollars. Column [1] derived at the Urban Institute.
The Role of UI as An Automatic Page 48 July 2010 Stabilizer During a Recession
Two comments about the deviation in column [4] can be offered. First, note how the
deviation grows between 2008Q3 and 2009Q2. While the NBER placed the cyclical peak
in 2007Q4, the downward trajectory in real GDP gains momentum later, during the last
half of 2008. Second, the caveat about the derivation of stable growth path needs to be
repeated. This was projected at the Urban Institute and not derived from the
Economy.com model. The projection assumes the economy after 2007Q3 grows at a rate
of 0.6 percent each quarter. Readers should view the deviations in column [4] as
illustrative.
Column [5] shows the real GDP deviation when real regular UI benefits respond to the
recession compared to constant real UI benefits. This deviation grows throughout the
quarters of 2008 and 2009, reaches $93 billion in 2009Q4, and averages more than $90
billion during late 2009 and 2010.
Columns [6] and [7] in Table 4.1 focus on real regular UI benefits. The pre-recession
level (2007Q3) of $28.1 billion grows to $77.2 billion (nearly tripling) by 2009Q4. Real
regular benefits then decline to $66.3 billion by 2010Q2. Column [7] shows the increases
from the pre-recession level of $28.1 billion. This deviation reaches $49.2 billion in
2009Q4. The deviation still exceeds $30 billion in mid-2010.
Column [8] shows the ratio of the real GDP deviation attributable to UI benefits (column
[2] less column [3] or column [5]) to the deviation in real UI benefits (column [7]). This
can be interpreted as the multiplier effect of UI benefits. For most periods, this ratio
ranges between 1.7 and 2.5. It shows the real GDP increment associated with each added
real dollar of regular UI benefits.
Note that these “multiplier” estimates are very large in the earliest periods, e.g., 4.5 in
2007Q4, but then decline to the more plausible 1.7 to 2.5 range starting in 2008Q3. Note
also that these estimates are based upon two simulated time paths from the 2009Q2
version of the Economy.com model. Readers are reminded that the results displayed in
Table 4.1 are built up from state-level detail.
The Role of UI as An Automatic Page 49 July 2010 Stabilizer During a Recession
The bottom row of Table 4.1 summarizes results for the eight calendar quarters from
2008Q3 to 2010Q2. The average downward deviation of real GDP averages $674 billion
while the increment to real GDP associated with increased UI benefits averages $71
billion. The UI multiplier effect on real GDP averages 2.0 and the share of the downward
deviation in real GDP filled by responsive UI benefits averages 0.11.
One exploration into the linkage between UI benefits and aggregate demand is to trace
the evolution of real disposable income of households (RYD). This series was traced for
the two simulations summarized in columns [2] and [3] of Table 4.1. When real UI
benefits respond to the recession, RYD is also noticeably higher compared to RYD when
real benefits are constant. The time paths of the deviations in RYD from these two
simulations strongly resemble the real GDP deviations shown in column [5] of Table 4.1.
The RYD deviations grow from $7 billion in 2007Q4 to $128 billion in 2009Q4,
averaging $122 billion during 2009Q3-2010Q2. For the same period real regular UI
benefits were higher by an average of $44 billion. The increase in real UI benefits
accounted for more than one-third of the increment in RYD.
To summarize, when real regular UI benefits respond to the recession they raise the level
of real GDP measurably above the level when real benefits are constant. During the eight
quarters of 2008Q3-2010Q2, real UI benefits cause an increment to real GDP that
averaged $71 billion and reduced by about 11 percent the downward deviation in real
GDP that would have occurred had real UI benefits not responded. The multiplier effect
of increased real benefits on real GDP averaged 2.0 during these eight quarters.
Since recipiency rates vary widely across states, it is relevant to examine the differing
state-level effects of UI on real GDP. Table 4.2 summarizes and contrasts selected
estimates from the 10 high-recipiency and 10 low-recipiency states. This state-level
analysis with 2009Q2 Economy.com model uses selected details from individual states
that underlie the national aggregates summarized in Table 4.1. The analysis emphasizes
the effects of real regular UI benefits on real GDP and does not attempt to project real
GDP under a no-recession scenario.
The Role of UI as An Automatic Page 50 July 2010 Stabilizer During a Recession
Table 4.2 summarizes these results with details in Panel A for the high-recipiency group
and Panel B for the low-recipiency group. Four features of Table 4.2 are noteworthy.
First, states in the low-recipiency group are on average larger. Their combined real GDP
is 52 percent larger.26
Second, both groups reach their real output trough in 2009Q2.
Third, the level of real UI benefits shows a greater contrast in 2007 than in 2010. The
negative feedback from the lagged TUR (due to exhaustions and lower monetary
eligibility) is larger on average in the high-recipiency states. Thus, aggregate real benefits
in the low recipiency group are less than half of real benefits in the high-recipiency group
in late 2007, but the proportion increases to 0.70 by early 2010. The increase in real
benefits after 2007Q3 averages $7.26 billion and $5.91 billion for the two groups
respectively. Fourth, the multipliers for regular UI benefits are similar across the two
groups of states and average 1.9 and 2.0, respectively. These multipliers are similar to the
national multipliers estimated earlier in Table 4.1. The principal conclusion of this state-
level analysis is that the cyclical performance of regular UI benefit payouts exhibits a
smaller contrast than the contrast in average recipiency rates discussed in Chapter 2.
26 Their employment is 59 percent larger.
The Role of UI as An Automatic Page 51 July 2010 Stabilizer During a Recession
Table 4.2. Real GDP and Real UI Benefits in High and Low Recipiency States, 2007Q3 to 2010Q2
Real GDP, Regular
Responds [1]
Real GDP,
Regular UI
Constant [2]
Real GDP, Responsive UI - Cons-
tant UI [1]-[2] [3]
Real Regular
UI Benefits
[4]
Change in Benefits
from 2007Q3
[5]
Real GDP Deviation/ Real
Ben. Deviation[3]/[5]
[6]
Panel A. 10 High-Recipiency States 2007Q3 1,714.0 1,714.0 0.0 6.89 - -
2007Q4 1,712.3 1,710.9 1.4 7.23 0.34 4.3
2008Q1 1,719.2 1,716.4 2.7 7.80 0.91 3.0
2008Q2 1,736.4 1,732.4 4.0 8.44 1.55 2.6
2008Q3 1,735.5 1,730.0 5.5 9.28 2.39 2.3
2008Q4 1,716.3 1,708.8 7.5 10.55 3.66 2.1
2009Q1 1,700.4 1,689.0 11.4 13.54 6.65 1.7
2009Q2 1,694.5 1,679.9 14.7 15.85 8.96 1.6
2009Q3 1,703.1 1,686.2 16.9 16.78 9.89 1.7
2009Q4 1,702.6 1,684.5 18.2 16.81 9.92 1.8
2010Q1 1,706.0 1,687.6 18.4 15.68 8.79 2.1
2010Q2 1,714.1 1,695.5 18.6 14.72 7.83 2.4 Average
08Q3-10Q2 1,709 1,695 13.9 14.15 7.26 1.9
Panel B. 10 Low-Recipiency States 2007Q3 2,596.8 2,596.8 0.0 3.19 - - 2007Q4 2,596.3 2,595.0 1.3 3.48 0.28 4.7 2008Q1 2,605.9 2,603.7 2.2 3.85 0.66 3.4 2008Q2 2,627.9 2,624.6 3.3 4.36 1.17 2.8 2008Q3 2,631.1 2,626.6 4.5 4.96 1.76 2.6 2008Q4 2,600.3 2,594.4 6.0 5.77 2.57 2.3 2009Q1 2,588.1 2,578.6 9.5 8.31 5.12 1.9 2009Q2 2,578.5 2,566.5 12.0 9.88 6.68 1.8 2009Q3 2,599.2 2,585.0 14.2 11.00 7.81 1.8 2009Q4 2,608.7 2,592.8 15.9 11.48 8.29 1.9 2010Q1 2,632.2 2,615.8 16.4 10.87 7.67 2.1 2010Q2 2,656.0 2,639.0 17.1 10.54 7.35 2.3
Average 08Q3-10Q2 2,612 2,600 11.9 9.10 5.91 2.0
Source: Simulations with the Economy.com model. Data measured in billions of 2000 dollars.
The Role of UI as An Automatic Page 52 July 2010 Stabilizer During a Recession
4.4 Extended UI Benefits
In December 2009, Emergency Unemployment Compensation or EUC beneficiaries
exceeded 4.0 million persons per week (more than 80 percent of the number receiving
regular UI benefits). By April 2010 EUC beneficiaries averaged 5.2 million and exceeded
the 4.6 million recipients of regular UI.
In April 2010, the Federal-State Extended Benefits (EB) program was providing
compensation in 38 states with weekly recipients averaging more than 0.2 million. This
number was only about one-third the number of EB recipients in October 2009. Roughly
half the states in 2009 enacted temporary triggers to activate EB based on the total
unemployment rate (TUR) from the household labor force survey. These temporary
triggers are slated to expire with the expiration of the stimulus package (American
Recovery and Reinvestment Act) at which time the number of active EB programs in the
states will likely decrease to fewer than 10.
Our simulations combine EUC and EB into a single extended benefits estimate. At the
time the simulations were specified, quarterly data existed from 2008Q3 through
2009Q3. For each state-quarter observation, EUC plus EB weeks compensated was
expressed as a proportion of regular UI weeks. Since weekly benefits for these programs
are based on regular UI weekly benefits, we assumed the increase in payments due to
extended benefits matches the proportional increase in weeks compensated from the two
extended benefit programs.
Other important assumptions for the combined extended benefits program were that the
proportions of weeks compensated for both EUC and EB in 2009Q3 were assumed to
hold during 2009Q4. Then a phase-down period was assumed during the first half of
2010. The combined extended benefit proportion in each state during 2010Q1 was
assumed to be two-thirds of its proportional size during 2009Q3, and one-third during
2010Q2. For the later quarters of 2010, the combined extended benefits proportion was
assumed to be zero. While this ignores the 11 states with permanent TUR triggers and
The Role of UI as An Automatic Page 53 July 2010 Stabilizer During a Recession
associated payments, these 11 states are generally small with modest aggregate
importance.27
The simulations also do not include the further extensions of EUC and EB
eligibility in 2009 and 2010 that were enacted between November 2009 and April 2010.
By assuming a phase-down in early 2010, the simulated payouts of both EUC and EB
understate actual payouts in 2010.
In examining the combined effects of EUC and EB, it should be noted that neither
program has exhibited a truly automatic response to higher unemployment as has the
regular UI program. Federal legislation in 2008, 2009, and 2010 created and then
extended the EUC program and greatly expanded the scope of EB. In contrast, the
increases in regular UI benefit payments have occurred automatically without any need
for legislation. While we acknowledge this distinction, all three aspects of UI have been
providing cash benefits to large numbers with unemployment. The simulations that
include EUC and EB benefits assume these benefits have the same kinds of effects on
household income and spending as regular UI benefits. For the two extended benefit
programs, the added payouts are assumed to occur between 2008Q3 and 2010Q2.
Extended benefits from EUC and EB combined are substantial. During the third and
fourth quarters of 2009 they were assumed to add some 74 to 75 percent to regular UI
benefit payments and some 49 to 50 percent during 2009Q2 and 2010Q1. Under the
statutes operative before November 2009, these benefits were scheduled to phase-out
during the first half of 2010.
Chart 4.5 adds combined extended benefits to the display shown previously in Chart 4.3.
As in Chart 4.3, all variables are measured in billions of 2000 dollars. The chart vividly
illustrates how responsive the combination of regular plus extended benefits was during
2009Q3 and 2009Q4 when their total approached $135 billion. The total increase in real
benefits for these two periods vis-a-vis 2007Q3 exceeds $100 billion. Note that taxes in
Chart 4.5 refer only to the state taxes that support regular UI benefits. In effect, the
27 The eleven are Alaska, Connecticut, Kansas, New Hampshire, New Jersey, New Mexico, North Carolina, Oregon, Rhode Island, Vermont, and Washington. In 2008 they represented 13.6 percent of UI employment.
The Role of UI as An Automatic Page 54 July 2010 Stabilizer During a Recession
simulations assume EUC and EB benefits are funded as a part of the overall federal
budget deficit.
Chart 4.5. Real UI Benefits and Taxes, 2007Q4 to 2013Q4 ($billions)
Source: Simulation results with the Economy.com model. Data in billions of 2000 dollars. To estimate the effects of extended benefits, the time paths of two simulations are
compared, one with just real regular UI benefits and one with total (regular plus
extended) UI benefits.28
Table 4.3 summarizes the findings with columns [1], [2] and [3]
showing real GDP and column [4] showing real combined extended benefits. Finally
column [5] again displays a multiplier, the response of real GDP to the payment of real
extended benefits.
Four features of Table 4.3 are noteworthy. First, extended benefits have measurable
effects on real GDP. The largest effects occur during 2009Q3 and 2009Q4, but the total
increase in real GDP averages more than $55 billion between 2008Q3 and 2010Q2.
Second, in the eight quarters when these benefits are paid (column [4]), they grow rapidly
28 Both simulations included UI taxes that respond to changes in regular UI benefits.
0
20
40
60
80
100
120
140
160
Real Regular UI Benefits Real UI Taxes Real Total (Regular plus Extended) UI Benefits
The Role of UI as An Automatic Page 55 July 2010 Stabilizer During a Recession
and then decrease rapidly. Note that the decrease in 2010 is partly a result of assumptions
about a phase-down made in late 2009 before the recent extensions of EUC and EB. Real
benefits during 2010Q1, for example, totaled $65 billion (in 2000 dollars measured at
annual rates) not $33.8 billion. Thus, real extended benefits are substantially understated
during 2010Q1 and 2010Q2.
Table 4.3. Real GDP and Real Extended Benefits, 2008Q1 to 2010Q2
Real GDP, With
Extended Benefits
[1]
Real GDP, No Extended Benefits
[2]
Real GDP,
Effect of Extended Benefits =
[1]-[2] [3]
Real Extended Benefits
[4]
Real GDP Deviation/Real Ext. Benefits
= [3]/[4] [5]
2008Q1 11,454 11,454 0.0 0.0 0.0
2008Q2 11,558 11,558 0.0 0.0 0.0
2008Q3 11,577 11,555 21.8 14.5 1.5
2008Q4 11,442 11,416 26.4 15.7 1.7
2009Q1 11,325 11,287 38.8 23.7 1.6
2009Q2 11,277 11,222 54.9 35.1 1.6
2009Q3 11,368 11,287 81.7 56.3 1.5
2009Q4 11,381 11,291 90.1 56.3 1.6
2010Q1 11,424 11,346 77.9 33.8 2.3
2010Q2 11,484 11,418 66.2 15.7 4.2 Average
08Q3-10Q2 11,410 11,353 57.2 31.4 2.0
Source: Simulation results with the Economy.com model. Data in billions of 2000 dollars.
Having a payment apparatus already in place (the administrative facilities of the state UI
programs) permits a rapid build-up and rapid decrease in extended UI benefits. Third, the
model suggests the multiplier for real extended benefits is somewhat smaller than for
regular UI benefits, but the difference is modest. The average in Table 4.1 was estimated
to be 2.0 and in Table 4.3 it is also 2.0. While the smallest “multiplier” in column [8] of
Table 4.1 is 1.7, the smallest multiplier in column [5] of Table 4.3 is 1.5.
Extended benefits provide an important addition to total benefit payments in all states
during the 2009Q2-2010Q1 period. Total simulated EUC plus EB payouts represented 61
percent of regular UI payments during these four quarters. For the 10 high-recipiency and
The Role of UI as An Automatic Page 56 July 2010 Stabilizer During a Recession
the 10 low-recipiency states the corresponding increases were 59 and 67 percent,
respectively. Measurable additions to disposable income, especially during these four
quarters, were present in all states.
To summarize, the payment of extended benefits has helped to sustain real GDP during
the “great recession” and estimates from the model suggest a per-dollar effect on real
GDP is about the same as the effect of regular UI benefits. The positive effect of
extended benefits during 2008Q3-2010Q2 raised real GDP by an average $57 billion per
quarter while regular program benefits raised real GDP by $71 billion over the same
period. Regular and extended benefits both operated to cushion the falloff in real GDP.
4.5 The Effects of UI Taxes
The regular UI programs in the states are financed with employer payroll taxes. Over
long periods these taxes roughly match regular UI benefit payments. Between 1990 and
2008 regular UI benefits and UI taxes averaged about 0.75 percent of the payroll of
taxable employers. While UI benefits directly increase household disposable income, UI
taxes add to costs for covered employers.29
In the Economy.com model UI taxes add to
the cost of doing business and reduce real output and employment.
The approach for estimating the effects of UI payroll taxes is to compare two simulated
run streams. The first simulates real GDP and other macro variables when UI benefits
and taxes respond to an increase in recession-related regular UI benefit payments. The
second simulates variables when benefits respond to the recession but UI taxes remain
constant in real terms. In the first of this pair of simulations, UI taxes respond with a lag
as summarized previously in Charts 4.3 and 4.5, and described in Chapter 2. In the
second, UI taxes only grow as the GDP deflator increases.
29 The effect of UI payroll taxes on employer labor costs involves an issue of tax incidence. To the extent that employers can shift the burden backward onto money wages, the actual incidence (or tax burden) falls on covered employees. Regardless of the incidence of the tax, UI financing imposes costs and offsets some or all of the positive effects of UI benefit payments.
The Role of UI as An Automatic Page 57 July 2010 Stabilizer During a Recession
Table 4.4 summarizes details on real GDP and real UI taxes for the 4 years 2007 to 2010.
Columns [1] and [2] respectively display model estimates of real GDP with real UI taxes
responsive to the recession and real taxes constant. Column [3] shows the difference, an
estimate of the effect on real GDP when UI taxes respond. Note in 2007 and 2008 the
effect of UI taxes is positive indicating that real GDP was slightly higher when real taxes
declined (rather than being held constant). Taxes decreased slightly in 2007 and 2008 in
response to earlier financing developments. Recall that the tax rate functions in the model
have current year taxes determined by a 4-year lag on benefits. Columns [4] and [5] next
display two tax series, respectively taxes responsive to higher benefit payouts and
constant real taxes and their difference in column [6]. Note the relatively long tail on the
tax response. A measurable tax response is first observed in 2009Q4, and the effect on
real GDP first exceeds $10 billion in 2010Q1. The aggregate time profile of the tax
response was displayed previously in Charts 4.3 and 4.5. Note also in Table 4.4 that the
average tax multiplier (-1.4 in column [7]) is smaller in absolute value than the average
multipliers for regular UI (2.0) and extended benefits (2.0).
The long, 4-year lag on the tax response means that the short run effect of the UI program
during a recession operates almost totally through increases in benefit payouts. Thus the
offsetting contractionary effects of taxes typically occur after the economy has started to
rebound. In the simulations summarized here, real UI taxes exceed $40 billion in all
quarters from 2010Q2 through 2015Q1. They reach a peak in 2012Q2, nearly three full
years after the peak in benefit payouts of 2009Q3 and 2009Q4.
The Role of UI as An Automatic Page 58 July 2010 Stabilizer During a Recession
Table 4.4. The Effect of UI Taxes on Real GDP, 2007Q1 to 2010Q2
Real GDP Reg Ben. & Taxes Respond
[1]
Real GDP, Reg Ben. but
Tax Constant
Real GDP Effect of
Higher UI Taxes=[1]-
[2][3]
Real UI Taxes
Respond [4]
Real UI Taxes
Constant [5]
Change in Real
Regular UI Taxes=[4]-
[5][6]
Real GDP Deviation/Real Tax Deviation=[3]/[6][7]
2007Q1 11,424 11,424 0 25.8 25.8 0.0 -
2007Q2 11,370 11,370 0 26.0 26.0 0.0 -
2007Q3 11,434 11,434 0 26.2 26.2 0.0 -
2007Q4 11,432 11,432 0 25.1 26.0 -0.84 -0.5
2008Q1 11,454 11,453 1 24.3 26.0 -1.68 -0.5
2008Q2 11,558 11,557 1 23.8 26.2 -2.32 -0.6
2008Q3 11,555 11,553 2 23.2 26.0 -2.77 -0.6
2008Q4 11,416 11,414 2 24.6 25.8 -1.22 -1.3
2009Q1 11,287 11,285 2 24.9 25.7 -0.80 -2.1
2009Q2 11,222 11,222 0 25.9 25.9 -0.01 -
2009Q3 11,287 11,289 -3 27.2 26.1 1.10 -2.5
2009Q4 11,291 11,300 -9 31.3 26.2 5.10 -1.8
2010Q1 11,346 11,361 -15 36.9 26.4 10.49 -1.4
2010Q2 11,418 11,434 -15 41.0 26.6 14.46 -1.1
2008Q3 -
2010Q2Av. 11,353 11,357 -5 29.4 26.1 3.29 -1.4 Source: Simulations with the Economy.com model. Data in billions of 2000 dollars.
4.6 The Net Effect of the UI Program
The net effect of UI on real GDP and other macro variables is the sum of three
components: The effect of regular UI benefits, the effect of extended benefits, and the
effect of UI taxes. This chapter used the Economy.com model to explore each of these
three components.
Table 4.5 summarizes the findings. Column [1] shows a projected time series of real GDP
with constant benefits and taxes that extends to 2010Q2. Columns [2], [3] and [4] then
show estimated effects of regular UI benefits, extended benefits (EUC plus EB) and UI
taxes respectively. Note that the effects of the financing of EUC and EB are not
considered in this analysis. Column [5] adds the three effects to yield a total estimated
effect of UI. The bottom line in Table 4.5 shows averages during 2008Q3-2010Q2.
The Role of UI as An Automatic Page 59 July 2010 Stabilizer During a Recession
The UI tax and benefit provisions added to the Economy.com model respond to the
recession as anticipated. Large increases in both regular UI benefits and extended
benefits were simulated. Charts 4.3 and 4.5 show a strong lagged response of UI taxes
following the “great recession”. Maximum real tax revenue is achieved in 2012Q2, three
years after the trough of the recession. The UI relationships included in the model
accurately track the actual patterns of UI benefits and taxes.
The behavioral relations in the model are state-level relations. The state-level findings
related to UI benefits and taxes are plausible and yielded one surprise. Recipiency rates in
the regular UI program vary widely across states. Comparisons of the 10 with highest
recipiency with the 10 with lowest recipiency showed that the former group had a
recipiency rate more than twice that of the low recipiency group. The respective
recipiency rate averages in 2007 were 0.47 and 0.19, more than a 2 to 1 ratio. The
differences in recipiency rates did not translate into comparable differences in stabilizing
effects. This high-low differential was closer to 1.5 to 1, whereas the recipiency rate
differential was more than 2.0 to 1. The explanation appears to be the stronger negative
feedback of lagged unemployment in the high-recipiency states. This finding should be
pursued with additional analysis.
The Role of UI as An Automatic Page 60 July 2010 Stabilizer During a Recession
Table 4.5. Net Effect of UI Program on Real GDP, 2007Q1 to 2010Q2
Real GDP, UI Benefits & UI Taxes
Constant [1]
Real GDP, Effect of Regular
UI Benefits [2]
Real GDP, Effect of Extended
UI Benefits [3]
Real GDP, Effect of UI Taxes
[4]
Net Effect of UI
Program = [2]+[3]+[4]
[5]
2007Q1 11,424 0 0 0 0
2007Q2 11,370 0 0 0 0
2007Q3 11,434 0 0 0 0
2007Q4 11,425 7 0 0 7
2008Q1 11,441 12 0 1 13
2008Q2 11,537 20 0 1 21
2008Q3 11,526 27 22 2 51
2008Q4 11,377 37 26 2 65
2009Q1 11,226 59 39 2 99
2009Q2 11,148 74 55 0 129
2009Q3 11,203 86 82 -3 165
2009Q4 11,207 93 90 -9 174
2010Q1 11,267 94 78 -15 157
2010Q2 11,338 95 66 -15 146
2008Q3 -
2010Q2 Avg. 11,287 71 57 -5 123 Source: Simulations with the Economy.com model. Data in billions of 2000 dollars.
Three features of Table 4.5 seem especially noteworthy. First, in this recession, extended
UI benefits play an important role in stabilizing real GDP. Their effect on real GDP
during 2008Q3-2010Q2 was $57 billion compared to an average of $71 billion for the
regular UI benefits. Second, all three aspects of the UI program affect real GDP. Given
the lags in the financing of regular UI benefits, however, the negative effects of UI taxes
commence only in late 2009 and peak only in 2012. After the onset of a recession, regular
UI financing does not immediately offset the positive effects of UI benefits. Third, the
combined effects of regular UI and extended benefits are substantial. During 2008Q3 -
2010Q2 their combined stimulative effects average $123 billion or more than one percent
of real GDP.30
30 The estimates derived from the Economy.com model refer to the marginal effect of increased UI benefits. Recall that $28.1 billion of real regular UI benefits do not enter the estimates summarized in Table 4.5.
The Role of UI as An Automatic Page 61 July 2010 Stabilizer During a Recession
4.7 The Stabilizing Effect of Unemployment Insurance
The simulations with the Economy.com model show a clear stabilizing effect of UI
benefits. The following paragraphs address some basic questions about the program’s
performance as an automatic stabilizer. It presents two sets of estimates of the stabilizing
effect of UI. Both can be used to describe the stabilizing effect. But, first there is a prior
question of what indicator should be used to assess the program’s stabilizing impact.
Previous research supported by the U.S. Department of Labor (Chimerine, et. al. (1999)
and Dunson, et.al. (1991)) focused on employment as well as real GDP. A review of
these two aggregates during the present downturn shows that they have not followed
identical time paths. Chart 4.6 traces real GDP and total employment from 2007Q3 to
2010Q4. All series are indexed at 100.0 in 2007Q3. The no-recession projection (made at
the Urban Institute) discussed earlier in Table 4.1 is shown as real GDP Index 0. The
next three GDP series are respectively: Real GDP with real taxes and benefits held
constant (Index 1), real GDP with regular benefits and UI taxes responding (Index 2), and
real GDP with real regular benefits, extended benefits (EUC plus EB) and taxes
responding (Index 3). Chart 4.6 also depicts three employment indices with the same
references to UI, e.g., Index 3 has regular benefits, extended benefits and UI taxes all
responding. Since the tax response is delayed into late 2009, the deviations of the Index 2
and Index 3 series from the Index 1 series is almost totally the effect of UI benefits.
Chart 4.6 provides a convenient summary of the scale of the recession in the deviation
between the three cyclical GDP series and the steady growth series. Under steady growth
of 0.6 percent per quarter, the real GDP Index 0 series reaches 108 by 2010Q4, or 6.0 full
index points above the highest of the 3 cyclical real GDP series.31
31 The analogues index for 2010Q4 is 111 under the more optimistic 2007Q4 Economy.com model.
The Role of UI as An Automatic Page 62 July 2010 Stabilizer During a Recession
Chart 4.6. Alternative Indices of Real GDP and Employment, 2007Q3 to 2010Q4
Source: Based data on simulations with the Economy.com model. All series equal 100.0 in 2007Q3.
A second obvious feature in Chart 4.6 is the sharp contrast in the time profiles of the
three recessionary real GDP projections versus the three employment projections. Real
GDP increases modestly after 2007Q3 and does not turn down until 2008Q4. All three
real output series reach a trough in 2009Q2 and then start to recover. By 2010Q4, all
three cyclical real GDP indices exceed 100. Real output has returned to its pre-recession
level. In contrast, employment starts to decrease in late 2008 but does not reach its trough
until the first and second quarters of 2010. The contrasting patterns of real output and
employment probably are specific to the “great recession” of 2008-2009. Other
recessions would likely have real output and employment patterns that are more closely
parallel. Because real GDP and employment have quite different time profiles, measures
of the stabilizing effect of UI could yield different results in an analysis that emphasizes
both measures. This analysis focuses on the time path of real GDP.
94.0096.0098.00
100.00102.00104.00106.00108.00110.00
Real GDP Index 0 Real GDP Index 1 Real GDP Index 2Real GDP Index 3 Emp. Index 1 Emp. Index 2Emp. Index 3
The Role of UI as An Automatic Page 63 July 2010 Stabilizer During a Recession
A third feature of Chart 4.6 is the clear effects of UI on real output and employment. The
regular UI program has a positive effect and extended benefits have a measurable
additional effect.32
How should these effects be described?
To measure the stabilizing effect of the UI program, at least two measures could be
considered.
Measure 1. Calculate the total deviation (shortfall) of actual GDP from steady
growth GDP. Note in Chart 4.6 that such measures can be calculated for each
calendar quarter after the onset of the recession as well as the average over a
longer period, e.g., 2008Q3-2010Q2. The effects of UI estimated from such
measures are proportions of the gap between steady growth path and no-UI-
program path that is closed by UI.
Measure 2. Calculate a peak-to-trough change in real GDP for two periods and
calculate the effects of UI on real output for the same two periods. The effects on
GDP due to UI will be some proportion of the change in real GDP between the
two periods.
Panel A of Table 4.6 displays a series of estimated gap-closing proportions based on the
results from Tables 4.1 and 4.5. The steady growth path is the same path that appears in
column [1] of Table 4.1, steady growth of 0.6 percent per quarter. Panel A uses this
steady growth series to estimate the downward deviation from potential when real UI
benefits and taxes are held constant (column [4] in Table 4.1). The gap-closing
proportions for the full UI program are shown in column [9]. These range between 0.094
and 0.273 (the quarter EUC started), and averaged 0.183 for the 2008Q3-2009Q2 period.
Note that both regular UI benefits and extended benefits contribute important elements in
closing the gap while UI taxes are unimportant.
32 While the text of the report emphasizes the effects of the UI program on real GDP, simulated effects on employment can also be noted. In 2009Q2, the trough quarter, real regular UI benefits raised total employment by 1.050 million while extended benefits caused an additional employment increase of 0.748 million and UI taxes had a negligible effect (a reduction of 0.002 million). During the eight quarters from 2008Q3 to 2010Q2, the estimated average effects on employment were real regular UI benefits (+0.891 million), real extended benefits (+0.714 million), and real UI taxes (-0.015 million).
The Role of UI as An Automatic Page 64 July 2010 Stabilizer During a Recession
The peak-to-trough calculations in Panel B show larger proportional gap-closing effects
of UI. This exercise obviously depends upon the choice of the peak and the trough. It was
previously noted that all real GDP series reached their trough in 2009Q2. The peak
selected was 2008Q2 because real GDP was higher than in the NBER-established peak of
2007Q4. In the Panel B comparison, the UI program is estimated to close 0.277 of the
gap with regular benefits and extended benefits making equal contributions.
In general, a peak-to-trough comparison of the type displayed in Panel B would be
expected to yield larger estimates of the gap-closing effects of the UI program. The
reason for this is that the peak-to-trough calculation would presumably use actual GDP
for two historic periods as peak and trough. This selection would omit the growth in
potential GDP between the peak period and the trough period. The omitted growth factor
would be larger as the time interval between the peak and the trough is longer. Note in
Table 4.6 that the decline in real GDP (column [1]) is estimated at $391, while in Panel A
the deviation from steady growth increased from $104 billion in 2008Q2 to $775 billion
in 2009Q2, an increase of $671 billion. The understatement of the loss of real GDP ($391
billion in Panel B versus $671 billion) would generally lead to an overstatement of the
proportional stabilization provided by the UI program.
Linking the preceding to earlier literature, two final comments can be offered. First, the
concern expressed by Dunson, et al. about the declining importance of UI as a stabilizer
does not extend to the “great recession” of 2008-2009. Early intervention with expansive
EUC and EB caused these extended benefits to add a large element to the stabilization
effect of UI. Second, as signaled by the real GDP and the employment projections of
Chart 4.6, the labor market of 2010 continues to have very high unemployment. The
annual TUR for 2010 may well exceed the 9.3 percent TUR of 2009. In this environment,
there will be continuing pressures to provide extended benefits to exhaustees. It is likely
that extended benefits will continue to rival in importance regular UI benefits as a
stabilizing element of the UI program.
The Role of UI as An Automatic Page 65 July 2010 Stabilizer During a Recession
Extended benefit payments (EUC plus EB) in 2009 totaled $49 billion and represented
0.35 percent of GDP. Across the span of 53 separate years that extend back to 1958, there
was an extended benefits program active in at least part of 28 separate years. For these 28
years the extended benefits-to-GDP percentage was highest in 2009. Extended benefit
programs in 1975 and 1976 had next-highest percentages at 0.28 percent of real GDP in
both years (recall Table 1.1). The EUC and EB programs have continued to receive
extensions in 2010, and, at least through April 2010, have continued to serve more than 5
million claimants per week. Consequently, a large gap-filling effect of the UI program
can be anticipated for 2010 with both regular UI and extended benefits being important.
The Role of UI as An Automatic Page 66 July 2010 Stabilizer During a Recession
Table 4.6. Summary: Estimated Stabilizing Effect of Unemployment Insurance on Real GDP
Deviation from
Steady Growth
[1]
---------------- Effect on Real GDP of ----------------- ------------Proportion of Gap Closed by ----------
Regular UI
Benefits [2]
Extended Benefits
[3]
UI Taxes [4]
Total UI Program [2+3+4]
[5]
Regular UI Benefits
[2]/[1] [6]
Extended Benefits [3]/[1]
[7]
UI Taxes [4]/[1]
[8]
Total UI Program
[5]/[1] [9]
Panel A. Estimated Effects on Real GDP by Calendar Quarter 2007Q4 78 6.92 - 0.42 7.34 0.089 - 0.005 0.094
2008Q1 131 12.44 - 0.91 13.35 0.095 - 0.007 0.102
2008Q2 104 19.53 - 1.34 20.87 0.187 - 0.013 0.200
2008Q3 185 27.19 21.82 1.66 50.67 0.147 0.118 0.009 0.273
2008Q4 404 36.74 26.40 1.54 64.68 0.091 0.065 0.004 0.160
2009Q1 626 58.56 38.80 1.69 99.05 0.094 0.062 0.003 0.158
2009Q2 775 74.29 54.89 -0.01 129.17 0.096 0.071 0.000 0.167
2009Q3 792 86.15 81.68 -2.70 165.13 0.109 0.103 -0.003 0.209
2009Q4 860 93.46 90.11 -9.23 174.34 0.109 0.105 -0.011 0.203
2010Q1 873 93.86 77.90 -14.56 157.20 0.108 0.089 -0.017 0.180
2010Q2 874 95.36 66.21 -15.44 146.13 0.109 0.076 -0.018 0.167
2008Q3 -
2010Q2 Av. 674 70.70 57.23 -4.63 123.30 0.105 0.085 -0.007 0.183 Panel B. Estimated Effects on Real GDP – Peak-to-Trough
Peak 08Q2 11,537 19.53 - 1.34 20.87
Trough 09Q2 11,146 74.29 54.89 -0.01 129.17 Change -391 54.76 54.89 -1.35 108.3 0.140 0.140 -0.003 0.277
Source: Based on simulations with the Economy.com model. Real GDP in billions of 2000 dollars.
The Role of UI as An Automatic Page 67 July 2010 Stabilizer During a Recession
4.8 Summary
This chapter has described the results of simulations with the Economy.com model.
Several conclusions can be drawn from the analysis. First, state-level detail regarding
regular UI benefits, extended benefits and UI taxes was successfully added to the
Economy.com model. State as well as national estimates of UI benefits and taxes were
developed and the resulting summary statistics were plausible. Benefits respond strongly
to increased unemployment and UI taxes respond strongly (but with a long lag) to
increases in regular UI benefit payouts. National summaries and summaries from 10
high-recipiency and 10 low-recipiency states showed that state-level UI variables were
successfully added to the Economy.com forecasting model.
The simulations that explored the effects of the UI program on the macro economy
yielded plausible results. The regular UI program provided measurable gap-filling
stabilization to the economy during 2008-2010. Real regular UI benefits reduced the
decline in real GDP during 2008Q3-2010Q2 by 0.105. Extended benefits had a slightly
smaller effect with a gap-filling proportion of 0.085 for the same period. Due to lags in
experience rating, the tax responses were considerably delayed. Chart 4.3 provided a
good visual summary of the lagged tax response. Through the second quarter of 2010, the
offsetting effects of UI taxes were small, but they will assume increasing importance in
years after 2010.
The results of the simulations presented in this paper suggest the following:
1) The size of the stabilizing effect of UI during 2007-2010 was larger than
found in previous research. This is partly due to the unusually large scale of
extended benefits payouts in the “great recession”.
2) The feasibility of conducting analysis at the state level is supported by the
findings. Somewhat surprisingly the stabilizing effects of UI in 10 low-
recipiency states was estimated to be about 70 percent of the stabilizing
The Role of UI as An Automatic Page 68 July 2010 Stabilizer During a Recession
effects in 10 high-recipiency states. The surprising finding is that a relative
stabilizing effect of 70 percent occurred even though the underlying
recipiency rate in low-recipiency states was less than half of that in high-
recipiency states, e.g., respective pre-recession recipiency rates of 0.19 versus
0.47. Stronger negative feedback from lagged unemployment is present in the
high-recipiency states so that recipiency increases less and decreases faster as
states go through the recession when compared to the low-recipiency states.
3) The per-dollar effects of UI taxes have been presumed to be smaller than the
effects of UI benefits. The simulations of this project supported this
presumption with the average tax multiplier estimated to be -1.4 compared to
2.0 for regular UI benefits, and 2.0 for extended benefits.
The Role of UI as An Automatic Page 69 July 2010 Stabilizer During a Recession
CHAPTER 5. CONCLUSION
A primary objective of the state unemployment insurance (UI) program is to provide
automatic or built-in stability to the macro economy. The present project has used the
Economy.com model to examine the performance of UI as an automatic stabilizer. The
analysis was conducted for individual states with national estimates derived by summing
the results from 51 separate state economies and UI programs.
The analysis developed state-level detail to describe UI benefit payments and taxes.
Benefit payments were estimated based on regression equations that described the
recipiency rates and replacement rates for the regular UI program in each state. Taxes
were estimated using state-level regression equations to explain average tax rates as a
percent of total UI-covered payroll. Proportional adjustments were then applied to the
average statewide tax rate to estimate tax rates for 19 detailed industries. Chapter 2 and
Appendix A provided details of the state-level benefit and tax relationships.
These relationships were imbedded into the Economy.com model. Details of the model
were given in Chapter 3 and Appendix B. The enhanced model was then used to simulate
macroeconomic performance and the stabilizing role of the state UI program during the
“great recession” of 2008-2009. Simulations were undertaken that yielded state and
national detail. While the simulations extended to 2020Q4, primary attention focused on
economic performance during 2007-2010. Chapter 4 summarized the results of the
simulations.
The simulations yielded two sets of conclusions. First, the behavioral relations describing
UI benefits and taxes yielded sensible findings about the response of benefits and taxes to
the recession. National summaries showed a large response of regular UI benefits,
extended benefits, and UI taxes. The tax response occurred with a long (4-year) lag with
The Role of UI as An Automatic Page 70 July 2010 Stabilizer During a Recession
effects that extended over several years following the cyclical peak in benefit payments
of 2009-2010. In fact, maximum tax revenues occurred in 2012Q2.
Second, the state-level detail built into the simulation model allowed one to study the
contrast in the response of benefit payments in high-recipiency states relative to low-
recipiency states. Vivid contrasts in the scale of UI benefit payments relative to real GDP
were documented in Chapters 2 and 4. Primarily due to differences in state-level
recipiency rates, UI benefits constitute a much larger share of real GDP in some states
than in others. The state-level contrasts extend over regions with high recipiency
concentrated in New England and Middle Atlantic States (6 of the 10 with the highest
recipiency rates) and low recipiency concentrated in states in the South and Rocky
Mountains (8 of the 10 with the lowest recipiency rates). This analysis documented these
contrasts and embedded them into the Economy.com model
The average multiplier effects of real UI variables on real GDP were plausible and higher
for regular (2.0) and extended benefits (2.0) than for real UI taxes (-1.4). Given the long
lags in the tax response, measurable negative effects of increased UI taxes will extend
from 2010 into several later years.
The analysis of the stabilizing performance of the UI program yielded generally plausible
results. The stabilizing effect of the regular UI program was estimated to close about one-
tenth of the real GDP shortfall caused by the recession. Extended benefits also played an
important stabilizing role. Because of lags that reflect experience rating, the response of
UI taxes was delayed with little increase in UI taxes occurring in 2009 and 2010. For the
three separate components of UI, the proportional gap-closing effects of the program
during 2008Q3-2010Q2 were as follows: Increased regular UI benefits = 0.105, extended
benefits = 0.085, and increased UI taxes = -0.007. On average, the UI program closed
0.183 of the gap in real GDP caused by the recession. For this particular recession, the UI
program has provided stronger stabilization of real output than in many past recessions.
The Role of UI as An Automatic Page 71 July 2010 Stabilizer During a Recession
REFERENCES Blank, Rebecca and David Card. 1991. “Recent Trends in Insured and Uninsured
Unemployment: Is there an Explanation?” Quarterly Journal of Economics, Vol. 106, No. 4, (November), pp. 1157-1189.
Burtless, Gary and Daniel Saks. 1984. “The Decline of Insured Unemployment During
the 1980s,” (Washington D.C.: The Brookings Institute). Chimerine, Lawrence, Theodore Black and Lester Coffey. 1999. “Unemployment
Insurance as an Automatic Stabilizer: Evidence of Effectiveness Over Three Decades,” Unemployment Insurance Occasional Paper 99-8, (Washington, D.C.: U.S. Department of Labor, Employment and Training Administration).
Cochrane, Steven. 2006. “The Moody’s Economy.com U.S. State Economic Model
System,” Moody’s Regional Financial Review, (July), pp. 4-7. Corson, Walter and Walter Nicholson. 1988. “An Examination of Declining UI Claims
During the 1980s,” Unemployment Insurance Occasional Paper 88-3, (Washington, D.C.: U.S. Department of Labor, Employment and Training Administration).
Di Natale, Marisa and Sophia Koropeckyi. 2007. “Forecasting U.S. Labor Force
Participation,” Moody’s Regional Financial Review, (November), pp. 20-27. Dunson, Bruce H, S. Charles Maurice and Gerald P. Dyer, Jr. 1991. “The Cyclical
Effects of the Unemployment Insurance (UI) Program: Final Report,” Unemployment Insurance Occasional Paper 91-3, (Washington, D.C.: U.S. Department of Labor, Employment and Training Administration).
Eilbott, Peter. 1966. “The Effectiveness of Automatic Stabilizers,” American Economic
Review, Vol. 56, No. 3, pp.450-465. Gruber, Jonathan. 1997. “The Consumption Smoothing Benefits of Unemployment
Insurance,” American Economic Review, Vol. 87, No. 1, pp.192-205. Oaxaca, Ronald and Carol Taylor. 1986. “Simulating the Impacts of Economics
Programs in Urban Areas: The Case of Unemployment Insurance Benefits,” Journal of Urban Economics, Vol. 19 (January), pp. 23-46.
Vroman, Wayne. 1991. “The Decline in Unemployment Insurance Claims Activity in the
1980s,” Unemployment Insurance Occasional Paper 91-2, (Washington, D.C.: U.S. Department of Labor, Employment and Training Administration).
The Role of UI as An Automatic Page 72 July 2010 Stabilizer During a Recession
Vroman, Wayne. 2009. “Unemployment Insurance in the American Recovery and Reinvestment Act,” The Urban Institute, (March).
Wing, Kennard T., Thomas H. Pollak and Amy Blackwood. 2008. “The Nonprofit
Almanac 2008,” The Urban Institute.
The Role of UI as An Automatic Appendix A-1 May 2010 Stabilizer During a Recession
APPENDIX A.
STATE-LEVEL REGRESSIONS
This appendix summarizes state-level regressions that examined important UI-related
behavioral relationships. All three sets of regressions to be described here were fitted
using annual time series data for individual states. Table A.1 displays regressions that
explain the average UI tax rate (as a percent of total payroll of taxable employers) for the
period 1960 to 2007. Table A.2 displays regressions to explain the recipiency rate in the
regular UI program (the weekly number of regular UI beneficiaries (WB) as a ratio to
total unemployment (TU) from BLS-LAUS data) for the years 1967 to 2007. Table A.3
displays regressions to explain the “Handbook” replacement rate (average weekly
benefits divided by the average weekly wage of all covered employees) for the 1967-
2007 period. In each table, the absolute value of the t ratio appears to the right of each
coefficient.
The Role of UI as An Automatic Appendix A-2 May 2010 Stabilizer During A Recession
Table A.1. Regressions of UI Effective Tax Rates by State on Lagged Benefit Ratios, 1960 to 2007
Type of Mean Mean Sum State Exp. Const. t B Ratio t B Ratio t B Ratio t B Ratio t Adj. Std. Durbin Tax Ben. Benefit
Rating-a Ratio Lag 1 Ratio Lag 2 Ratio Lag 3 Ratio Lag 4 Ratio R2 Error Watson Rate% Ratio% Coeff.
ALABAMA BR-3 -0.302 3.8 0.438 4.2 0.243 2.2 0.364 3.5 0.260 2.9 0.826 0.176 0.76 0.803 0.802 1.304 ALASKA PD 0.570 2.0 0.210 1.4 0.144 0.7 0.155 0.8 0.282 2.0 0.418 0.384 0.33 2.194 1.943 0.789
ARIZONA RR -0.024 0.5 0.103 1.9 0.349 5.5 0.241 3.8 0.293 5.5 0.840 0.104 0.74 0.620 0.621 0.987
ARKANSAS RR 0.125 1.4 0.213 3.7 0.276 4.5 0.173 2.8 0.148 2.6 0.698 0.121 0.63 0.990 1.045 0.810
CALIFORNIA RR -0.187 2.9 0.293 4.7 0.493 7.0 0.145 2.1 0.176 3.1 0.910 0.126 0.80 1.126 1.162 1.107
COLORADO RR -0.020 0.3 0.327 3.1 0.406 3.1 0.192 1.5 0.042 0.4 0.722 0.136 0.93 0.566 0.592 0.966
CONNECTICUT BR-3 0.365 6.3 0.114 2.2 0.168 2.9 0.088 1.6 0.142 3.2 0.716 0.146 0.75 0.948 1.086 0.512
DELAWARE BWR 0.187 1.5 0.147 1.1 0.236 1.5 0.188 1.2 0.172 1.4 0.401 0.249 0.27 0.859 0.877 0.743
DIST OF COL RR 0.053 1.3 0.050 0.5 0.330 2.1 0.208 1.4 0.230 2.3 0.862 0.109 0.73 0.688 0.763 0.817
FLORIDA BR-3 -0.149 2.7 0.452 5.4 0.314 3.1 0.406 4.0 0.120 1.5 0.818 0.118 0.36 0.547 0.526 1.291
GEORGIA RR 0.242 6.5 0.088 1.8 0.206 3.8 0.171 3.2 0.134 2.9 0.860 0.087 0.46 0.580 0.579 0.598
HAWAII RR -0.039 0.3 0.351 2.5 0.316 1.5 0.110 0.5 0.261 1.9 0.664 0.232 0.85 1.154 1.143 1.038
IDAHO RR -0.190 1.4 0.300 2.6 0.291 2.1 0.208 1.5 0.296 2.6 0.695 0.231 0.43 1.222 1.242 1.094
ILLINIOS BR-3 -0.039 0.6 0.144 1.9 0.385 4.4 0.255 3.0 0.182 2.7 0.847 0.157 0.56 0.930 0.987 0.966
INDIANA RR 0.175 4.1 0.221 3.7 0.204 3.6 0.178 3.2 0.046 0.9 0.694 0.112 0.59 0.576 0.606 0.649
IOWA BR-3 -0.227 3.4 0.393 4.3 0.281 2.5 0.241 0.3 0.300 3.3 0.860 0.172 0.56 0.815 0.870 1.215
KANSAS RR 0.298 3.4 0.198 2.3 0.161 1.7 0.224 2.3 0.055 0.7 0.748 0.147 0.46 0.765 0.791 0.638
KENTUCKY RR 0.178 3.0 0.284 4.6 0.236 3.8 0.139 2.2 0.095 1.8 0.826 0.140 0.47 0.982 1.019 0.754
LOUISIANA RR 0.172 2.6 0.221 2.8 0.193 1.8 0.094 0.9 0.167 2.0 0.742 0.194 0.83 0.856 0.983 0.675
MAINE RR 0.255 2.5 0.242 2.8 0.164 1.8 0.156 1.8 0.178 2.4 0.667 0.203 0.61 1.192 1.207 0.741
MARYLAND BR-3 -0.277 3.0 0.133 1.1 0.490 3.8 0.394 3.2 0.249 2.6 0.815 0.209 0.71 0.872 0.844 1.267
MASSACHUSETTS RR 0.139 1.6 0.188 2.7 0.274 3.2 0.118 1.4 0.265 4.1 0.790 0.156 0.74 1.231 1.265 0.845
MICHIGAN BR-RR 0.331 4.1 0.066 1.3 0.201 4.2 0.200 4.3 0.232 5.4 0.772 0.187 0.92 1.244 1.254 0.699
MINNESOTA BR-4 0.215 2.6 0.082 0.9 0.217 2.2 0.179 1.9 0.210 2.6 0.592 0.152 0.32 0.860 0.911 0.689
MISSISSIPPI BR-3 0.045 0.5 0.199 1.5 0.391 2.5 0.383 2.5 -0.062 0.5 0.631 0.242 0.44 0.815 0.802 0.912
MISSOURI RR -0.155 2.6 0.266 4.2 0.270 3.9 0.328 4.8 0.235 3.9 0.824 0.097 1.46 0.686 0.759 1.098
MONTANA RR 0.499 4.8 0.368 2.3 -0.016 0.1 0.258 1.6 -0.131 1.2 0.467 0.227 0.38 1.060 1.118 0.479
NEBRASKA RR 0.006 0.2 0.585 8.4 0.153 1.8 0.132 1.6 0.057 0.8 0.833 0.084 1.23 0.552 0.578 0.927
NEVADA RR 0.018 0.2 0.259 2.4 0.219 1.9 0.229 2.1 0.260 2.9 0.706 0.205 0.80 1.146 1.101 0.967
NEW HAMPSHIRE RR 0.018 0.5 0.336 6.9 0.162 3.0 0.207 3.9 0.174 3.9 0.872 0.112 0.74 0.584 0.602 0.879
NEW JERSEY RR 0.241 2.1 0.182 1.8 0.111 0.9 0.154 1.3 0.295 3.2 0.652 0.199 0.82 1.307 1.390 0.742
NEW MEXICO RR 0.260 2.7 0.174 1.6 0.200 1.5 0.085 0.6 0.169 1.6 0.378 0.150 0.25 0.792 0.841 0.628
NEW YORK RR -0.095 1.4 0.277 3.6 0.242 2.8 0.224 2.8 0.257 4.0 0.869 0.121 0.85 1.010 1.054 0.999
The Role of UI as An Automatic Appendix A-3 May 2010 Stabilizer During A Recession
Table A.1. Regressions of UI Effective Tax Rates by State on Lagged Benefit Ratios, 1960 to 2007 (cont)
State Type of
Exp. Rating-a
Const. t Ratio
B Ratio Lag 1
t Ratio
B Ratio Lag 2
t Ratio
B Ratio Lag 3
t Ratio
B Ratio Lag 4
t Ratio Adj.
R2 Std.
Error Durbin Watson
Mean Tax
Rate%
Mean Ben.
Ratio%
Sum Benefit Coeff.
NORTH CAROLINA RR 0.034 0.5 0.125 1.9 0.264 3.9 0.264 3.9 0.229 3.8 0.748 0.149 0.61 0.701 0.715 0.883
NORTH DAKOTA RR 0.050 0.6 0.679 5.7 0.114 0.7 0.133 0.8 0.030 0.2 0.839 0.173 0.91 1.163 1.138 0.956
OHIO RR 0.171 3.2 0.059 1.1 0.206 4.0 0.187 3.7 0.253 5.6 0.810 0.147 0.88 0.842 0.913 0.705
OKLAHOMA BWR -0.117 2.0 0.384 4.6 0.371 3.9 0.345 3.6 0.034 0.4 0.822 0.123 0.62 0.645 0.644 1.134
OREGON BR-3 0.189 1.0 0.239 2.1 0.258 1.9 0.150 1.1 0.245 2.3 0.476 0.299 0.50 1.450 1.367 0.893
PENNSYLVANIA BR-RR 0.343 3.8 0.019 0.3 0.246 3.2 0.208 2.9 0.231 3.8 0.782 0.194 0.80 1.403 1.444 0.704
RHODE ISLAND RR 0.558 3.1 0.132 1.4 0.205 2.0 0.081 0.8 0.216 2.4 0.443 0.297 0.52 1.683 1.737 0.634
SOUTH CAROLINA RR 0.281 4.2 0.157 2.9 0.126 2.2 0.139 2.4 0.164 3.0 0.509 0.139 0.50 0.741 0.767 0.586
SOUTH DAKOTA RR -0.083 1.8 0.397 3.6 0.234 1.6 0.162 1.1 0.214 1.9 0.773 0.108 0.67 0.445 0.509 1.007
TENNESSEE RR 0.145 2.1 0.206 2.5 0.184 2.2 0.238 2.9 0.135 1.9 0.665 0.158 0.45 0.780 0.781 0.763
TEXAS BR-3 -0.083 1.4 0.535 4.3 0.419 2.9 0.212 1.5 -0.097 0.8 0.732 0.133 1.44 0.482 0.513 1.069
UTAH BR-4 -0.127 1.9 0.243 2.8 0.420 4.0 0.269 2.6 0.173 2.1 0.844 0.137 0.67 0.841 0.840 1.105
VERMONT BR-3 0.271 1.6 -0.170 1.4 0.238 1.7 0.252 1.8 0.369 3.3 0.497 0.300 0.38 1.162 1.273 0.689
VIRGINIA BR-4 -0.144 2.6 0.419 3.9 0.398 3.4 0.265 2.3 0.238 2.4 0.747 0.136 0.60 0.442 0.425 1.321
WASHINGTON BR-4 0.689 3.9 0.035 0.4 0.221 1.8 0.068 0.5 0.195 2.0 0.323 0.269 0.31 1.485 1.481 0.519
WEST VIRGINIA RR 0.208 2.9 0.294 3.8 0.253 3.0 0.080 1.0 0.115 1.8 0.808 0.195 1.30 1.124 1.195 0.743
WISCONSIN RR -0.022 0.2 -0.002 0.0 0.271 2.6 0.237 2.3 0.438 5.0 0.697 0.238 0.58 1.039 1.117 0.943
WYOMING BR-3 0.196 2.8 0.354 4.2 0.117 1.0 0.332 3.0 -0.018 0.2 0.764 0.220 0.45 0.940 0.913 0.785
Source: Regressions based on data in columns (15) and (16) of the "UI Financial Handbook," (1995) and subsequent Handbook updates. Absolute value of t ratios appear to the right of each coefficient. Regressions for Georgia and Kansas also included dummy variables for periods of UI tax holidays. a - RR - Reserve Ratio, BR - Benefit Ratio (and years of benefits), BWR - Benefit Wage Ratio , PD - Payroll Decline.
The Role of UI as An Automatic Appendix A-4 May 2010 Stabilizer During A Recession
Table A.2. UI Recipiency Rates, Time Series Regression Results for Individual States, 1967 to 2007 State
Constant t Ratio TUR t
Ratio TUR Lag
t Ratio
Dummy 1981
t Ratio
Dummy 1996
t Ratio
Adj. R2
Std. Error
Durbin Watson
Mean Recip. Rate
ALABAMA 0.273 11.9 0.980 1.8 -0.680 1.2 -0.060 3.4 0.074 3.6 0.372 0.037 1.36 0.273
ALASKA 0.953 5.0 -0.482 0.2 -3.397 1.5 -0.042 0.8 -0.080 1.2 0.015 0.144 0.39 0.571
ARIZONA 0.225 7.8 1.684 3.3 -1.884 3.7 -0.008 0.5 -0.007 0.4 0.252 0.039 1.52 0.206
ARKANSAS 0.283 6.2 1.984 2.5 -2.321 2.7 0.021 1.1 0.060 2.4 0.447 0.043 0.82 0.293
CALIFORNIA 0.453 22.2 0.935 2.7 -2.132 6.3 -0.003 0.4 -0.011 1.2 0.514 0.021 1.14 0.364
COLORADO 0.098 5.4 1.484 3.4 -0.405 0.9 0.043 4.0 0.004 0.3 0.567 0.024 1.17 0.181
CONNECTICUT 0.602 11.4 2.180 1.9 -4.203 3.8 -0.089 2.9 0.111 3.5 0.437 0.080 0.83 0.473
DELAWARE 0.396 9.9 1.232 1.2 -1.761 1.8 0.001 0.0 0.114 4.7 0.487 0.057 1.73 0.403
DIST OF COL 0.348 8.0 1.856 1.7 -1.274 1.2 -0.011 0.4 -0.036 1.4 0.076 0.066 1.33 0.371
FLORIDA 0.105 6.6 1.903 5.5 -1.020 3.1 -0.009 1.0 0.074 7.4 0.656 0.022 1.11 0.173
GEORGIA 0.131 3.7 4.610 5.1 -2.992 3.3 0.015 0.8 0.005 0.3 0.378 0.045 1.70 0.228
HAWAII 0.376 12.4 3.332 3.6 -3.488 3.9 0.031 1.7 -0.035 2.1 0.326 0.043 1.29 0.377
IDAHO 0.298 6.3 1.958 1.7 -2.285 1.8 0.055 2.7 0.047 1.9 0.414 0.046 0.90 0.328
ILLINIOS 0.313 12.4 2.640 4.2 -1.609 2.3 -0.100 4.6 0.089 4.4 0.544 0.040 1.26 0.337
INDIANA 0.226 14.0 2.264 5.3 -1.870 4.2 -0.051 3.9 0.101 7.1 0.704 0.031 1.26 0.245
IOWA 0.339 16.3 3.118 3.0 -3.367 3.0 -0.037 1.7 0.082 4.0 0.498 0.042 1.45 0.329
KANSAS 0.273 9.3 5.327 6.1 -4.217 4.7 -0.004 0.2 -0.051 3.7 0.587 0.034 1.28 0.305
KENTUCKY 0.282 7.8 2.119 2.7 -1.865 2.2 -0.050 1.8 0.039 1.5 0.236 0.048 0.97 0.277
LOUISIANA 0.167 4.9 2.810 4.0 -1.402 2.0 -0.055 3.0 0.026 1.1 0.388 0.039 1.25 0.241
MAINE 0.469 14.6 1.477 1.9 -2.442 3.3 -0.045 2.7 -0.041 2.2 0.434 0.044 1.15 0.374
MARYLAND 0.320 14.0 2.563 3.6 -3.128 4.4 -0.015 1.2 -0.012 0.9 0.379 0.031 1.12 0.279
MASSACHUSETTS 0.720 29.4 0.471 0.9 -4.412 8.4 -0.073 4.9 0.037 2.2 0.822 0.040 1.01 0.465
MICHIGAN 0.387 24.6 2.069 7.2 -2.757 9.1 -0.037 3.1 0.067 5.0 0.832 0.028 2.08 0.333
MINNESOTA 0.328 12.4 2.071 2.6 -1.594 1.9 -0.035 2.2 0.045 2.5 0.240 0.037 1.46 0.341
MISSISSIPPI 0.182 7.9 2.214 4.3 -1.990 3.7 -0.003 0.1 0.031 1.8 0.338 0.034 1.44 0.206
MISSOURI 0.428 16.5 1.806 2.6 -3.321 4.5 -0.049 2.6 -0.004 0.2 0.604 0.039 2.04 0.316
MONTANA 0.330 6.7 -1.180 0.9 -0.651 0.5 -0.015 0.8 0.036 1.4 0.102 0.046 0.45 0.300
NEBRASKA 0.253 11.7 1.396 1.6 -0.042 0.0 -0.020 1.5 0.000 0.1 0.082 0.035 1.23 0.285
NEVADA 0.397 11.3 2.380 3.7 -2.112 3.2 -0.101 7.1 0.060 3.3 0.611 0.038 1.18 0.364
NEW HAMPSHIRE 0.315 9.4 4.133 3.6 -3.648 3.2 -0.128 4.7 -0.005 0.2 0.540 0.068 1.85 0.251
NEW JERSEY 0.707 27.8 0.025 0.0 -2.909 4.9 -0.110 7.7 0.040 2.5 0.769 0.038 1.09 0.476
NEW MEXICO 0.296 7.9 1.012 1.3 -2.145 2.6 -0.010 0.7 -0.005 0.3 0.198 0.035 1.12 0.211
NEW YORK 0.641 31.3 0.281 0.6 -3.142 6.6 -0.074 6.9 -0.060 5.2 0.852 0.029 1.53 0.395
The Role of UI as An Automatic Appendix A-5 May 2010 Stabilizer During A Recession
Table A.2. UI Recipiency Rates, Time Series Regression Results for Individual States, 1967 to 2007 (cont)
Constant t Ratio
TUR t Ratio
TUR Lag
t Ratio
Dummy 1981
t Ratio
Dummy 1996
t Ratio
Adj. R2
Std. Error
Durbin Watson
Mean Recip. Rate State
NORTH CAROLINA
0.194 8.1 3.435 6.4 -2.120 3.9 -0.015 1.1 0.056 3.7 0.550 0.037 1.18 0.268
NORTH DAKOTA 0.112 1.5 2.570 1.3 2.093 1.0 -0.025 1.0 0.062 1.7 0.084 0.057 0.45 0.305
OHIO 0.222 10.7 2.931 6.4 -2.488 5.2 -0.001 0.1 0.022 1.3 0.497 0.035 1.00 0.258
OKLAHOMA 0.256 7.4 1.710 2.2 -1.872 2.4 -0.058 2.7 0.008 0.3 0.320 0.044 1.24 0.212
OREGON 0.433 15.0 1.457 2.4 -2.875 4.7 0.051 3.3 0.000 0.0 0.465 0.040 1.13 0.369
PENNSYLVANIA 0.416 16.2 3.191 4.8 -2.686 4.0 -0.058 3.4 0.107 5.8 0.591 0.040 0.75 0.441
RHODE ISLAND 0.792 24.6 -1.784 2.8 -1.672 2.7 -0.086 4.3 -0.076 3.5 0.689 0.054 1.46 0.512
SOUTH CAROLINA
0.192 5.4 3.489 4.7 -2.574 3.4 -0.008 0.4 0.034 1.7 0.341 0.049 1.55 0.251
SOUTH DAKOTA 0.115 4.7 2.760 2.8 0.323 0.3 -0.101 7.4 0.040 3.1 0.654 0.025 0.96 0.169
TENNESSEE 0.351 13.9 2.598 3.9 -3.333 4.9 -0.038 1.9 0.028 1.4 0.497 0.042 1.23 0.292
TEXAS 0.081 3.9 2.803 5.8 -1.429 3.0 0.003 0.2 0.044 4.1 0.661 0.021 1.41 0.174
UTAH 0.211 6.2 1.888 2.5 -0.595 0.7 -0.049 3.2 -0.002 0.0 0.407 0.040 0.75 0.245
VERMONT 0.438 18.5 1.476 2.4 -1.860 3.2 0.006 0.4 0.052 3.6 0.450 0.034 1.68 0.439
VIRGINIA 0.052 2.0 4.459 5.4 -2.118 2.6 0.000 0.0 0.089 5.3 0.572 0.034 1.25 0.180
WASHINGTON 0.437 9.0 1.959 2.2 -2.856 3.3 0.009 0.4 0.003 0.1 0.170 0.060 0.58 0.381
WEST VIRGINIA 0.289 7.1 2.003 2.9 -1.946 2.7 -0.054 1.8 0.061 1.7 0.279 0.053 0.40 0.274
WISCONSIN 0.404 16.8 1.963 2.8 -2.866 3.9 0.030 1.7 0.102 5.4 0.701 0.041 1.17 0.410
WYOMING 0.222 6.0 3.798 3.6 -4.515 4.4 0.103 3.7 -0.028 1.1 0.508 0.050 1.05 0.246
Source: Data on recipiency rates (WBTU) and unemployment rates (TURs) from OWS and BLS . Dummy variables equal 1.0 for the period 1981 to 2007 (Dummy 1981) and 1.0 for the period 1996 to 2007 (Dummy 1996), zero otherwise. Absolute values of t ratios appear to the right of each coefficient.
The Role of UI as An Automatic Appendix A-6 May 2010 Stabilizer During A Recession
Table A.3. Replacement Rate Regressions, 1967 to 2007
State Constant t Ratio
MaxBen/ AWW
t Ratio TUR t
Ratio TUR Lag
t Ratio
RRate Stat.
t Ratio
2 High Quarter Dummy
t Ratio
Annual Wage
Dummy
t Ratio Adj.
R2 Std.
Error Durbin Watson
Mean Repl. Rate
ALABAMA 0.132 9.7 0.485 16.5 0.274 2.7 -0.277 2.9 -0.015 5.5 0.930 0.007 1.24 0.307
ALASKA 0.103 6.8 0.428 16.5 0.226 2 0.884 0.011 1.10 0.254
ARIZONA 0.026 1 0.724 9 0.734 3.6 -0.292 1.4 0.772 0.016 0.50 0.312
ARKANSAS 0.240 12.2 0.294 11.1 0.127 0.7 -0.367 1.8 -0.017 3 0.796 0.010 1.91 0.389
CALIFORNIA -0.068 1.5 0.381 3.3 0.499 1.47 -0.109 0.3 0.407 2.9 0.721 0.022 0.19 0.292
COLORADO 0.194 5.5 0.460 6.1 0.281 1.3 -0.518 2.5 -0.029 7.5 0.781 0.012 1.25 0.403
CONNECTICUT 0.146 5.4 0.525 7.7 0.049 0.3 -0.449 2.7 -0.072 15.2 0.910 0.012 0.75 0.345
DELAWARE 0.099 2.6 0.522 5.9 1.096 2.5 -0.977 2.1 0.483 0.025 0.97 0.337
DIST OF COL -0.032 0.5 0.469 11.7 0.093 0.3 -0.581 2.3 0.341 2.5 0.940 0.016 1.02 0.337
FLORIDA 0.170 9.9 0.406 9.98 0.867 3.3 -0.776 3 -0.019 2.8 0.771 0.016 1.89 0.344
GEORGIA 0.120 5.4 0.535 9.6 0.458 2.5 -0.448 2.5 0.703 0.009 0.88 0.332
HAWAII -0.129 2.3 0.939 11.1 0.469 1.2 -0.467 1.2 0.760 0.018 1.17 0.467
IDAHO 0.290 14.5 0.208 5.7 0.707 3.8 -0.605 3 0.471 0.008 1.08 0.407
ILLINIOS -0.452 4.2 0.407 8.7 0.970 5.3 -0.450 2.3 1.246 6.3 0.845 0.011 0.94 0.356
INDIANA 0.131 14.9 0.508 29.9 0.521 3.8 -0.200 1.5 0.961 0.010 0.65 0.321
IOWA 0.101 4.9 0.641 15.1 1.216 6 -0.738 3.7 0.892 0.008 1.10 0.438
KANSAS 0.149 7.4 0.502 11.1 1.147 5 -0.900 3.7 0.847 0.009 1.12 0.424
KENTUCKY -0.003 0.1 0.531 9.8 0.617 4.5 -0.462 2.9 0.238 3.3 -0.057 9.3 0.909 0.009 1.57 0.368
LOUISIANA -0.269 4.1 0.629 11.3 0.394 1.3 -0.874 3.3 0.727 5.7 -0.043 5.8 0.874 0.019 0.68 0.348
MAINE -0.044 0.7 0.365 3.3 0.125 1 0.440 4.2 -0.012 2.1 0.554 0.010 0.52 0.389
MARYLAND 0.097 5.1 0.569 13.3 1.078 4.7 -0.856 3.7 0.838 0.010 0.99 0.356
MASSACHUSETTS 0.286 7.5 0.174 2.3 0.572 3 -0.473 2.5 0.232 0.014 0.67 0.381
MICHIGAN -0.098 1 0.336 11.1 0.775 5.4 -0.291 1.8 0.538 3.2 -0.040 3 0.813 0.013 1.90 0.362
MINNESOTA 0.161 11.9 0.447 17.2 0.259 1.9 0.897 0.011 0.41 0.413
MISSISSIPPI 0.181 11 0.361 8.6 0.130 1 -0.148 1.1 0.648 0.009 1.97 0.324
MISSOURI 0.063 1.9 0.560 7.7 0.558 2.8 -0.615 2.8 0.092 1.9 0.717 0.011 1.54 0.321
MONTANA 0.113 4.8 0.416 7.6 1.245 5.3 -0.019 1.6 0.799 0.017 1.40 0.391
NEBRASKA -0.268 4.2 0.681 11.6 0.529 2.8 -0.159 0.8 0.615 5 0.000 0 0.872 0.007 1.51 0.365
NEVADA 0.213 9.4 0.270 5.6 0.778 3.2 -0.234 1 0.542 0.014 1.43 0.365
NEW HAMPSHIRE 0.094 4.7 0.574 13.9 0.179 0.8 -0.505 2.2 0.835 0.014 0.97 0.341
NEW JERSEY 0.080 1.3 0.247 4.3 0.365 1.8 -0.550 2.9 0.282 4.1 0.397 0.013 0.66 0.364
The Role of UI as An Automatic Appendix A-7 May 2010 Stabilizer During A Recession
Table A.3. Replacement Rate Regressions, 1967 to 2007 (cont.)
State Constant t Ratio
MaxBen/ AWW
t Ratio TUR t
Ratio TUR Lag
t Ratio
RRate Stat.
t Ratio
2 High Quarter Dummy
t Ratio
Annual Wage
Dummy
t Ratio Adj.
R2 Std.
Error Durbin Watson
Mean Repl. Rate
NEW MEXICO -0.233 1.8 0.359 10.9 0.554 3.2 -0.364 2.1 0.810 3.2 0.789 0.008 1.24 0.353
NEW YORK 0.150 6.3 0.378 6.4 0.451 2 -0.748 3.4 0.028 5.1 0.677 0.013 0.61 0.307 NORTH CAROLINA 0.169 4 0.360 5.5 0.352 2.1 -0.350 2.1 -0.002 0.5 -0.017 1.4 0.890 0.011 1.46 0.368
NORTH DAKOTA 0.205 7.4 0.315 6.8 1.238 4.1 0.625 0.017 0.60 0.431
OHIO 0.177 7.1 0.347 6.1 0.862 6.1 0.662 0.019 0.99 0.376
OKLAHOMA -0.107 1.9 0.522 13.9 0.445 1.9 -0.343 1.5 0.414 3.5 0.926 0.013 1.46 0.372
OREGON 0.177 13.8 0.333 16.2 0.637 3.9 -0.450 2.8 0.869 0.011 0.94 0.369
PENNSYLVANIA 0.234 13 0.217 6.5 1.109 5 -0.386 1.7 0.762 0.014 1.04 0.406
RHODE ISLAND 0.244 4 0.315 3.2 0.436 2 -0.518 2.3 -0.003 0.3 0.571 0.019 0.73 0.420 SOUTH CAROLINA 0.199 18.6 0.340 15.5 0.287 2.8 -0.355 3.5 0.879 0.007 1.45 0.354
SOUTH DAKOTA -0.041 1.6 0.470 9.6 1.281 5.1 0.329 8.1 0.914 0.009 1.35 0.399
TENNESSEE 0.055 1.9 0.334 5.8 0.432 3 -0.397 2.7 0.258 5.7 0.787 0.009 0.87 0.315
TEXAS 0.092 8.7 0.535 14.7 1.222 6.4 -0.779 3.6 0.944 0.009 1.05 0.347
UTAH 0.104 3.1 0.510 9.4 0.561 3.3 0.688 0.015 0.86 0.416
VERMONT 0.207 13.2 0.389 11.8 0.621 3.9 -0.638 4.1 -0.015 4.9 0.822 0.009 1.05 0.393
VIRGINIA 0.088 4 0.592 9.6 0.531 1.5 -0.446 1.4 0.820 0.013 1.37 0.352
WASHINGTON 0.182 5.4 0.295 5.9 1.102 4.2 -0.570 2.1 0.007 0.7 -0.033 2.1 0.843 0.017 1.64 0.376
WEST VIRGINIA -0.019 6 0.439 16.2 0.627 3.8 -0.463 2.7 0.551 7.3 0.950 0.013 1.32 0.354
WISCONSIN 0.254 8.9 0.209 3.8 1.055 4.7 -0.557 2.5 0.017 3.8 0.671 0.013 1.06 0.405
WYOMING 0.117 6.1 0.538 11.6 0.656 3.1 -0.279 1.3 0.870 0.011 0.92 0.397 Source: Replacement rates, maximum benefits, average weekly wages and methods of calculating weekly benefits from the Office of Workforce Security. Data on TURs from BLS-LAUS program. Absolute value of t ratios appear to the right of each coefficient.
The Role of UI as An Automatic Appendix B-1 May 2010 Stabilizer During a Recession
APPENDIX B. COST OF DOING BUSINESS INDEX
The cost of doing business index estimates how business costs differ across states, within states,
and across industries. The index is a weighted sum of three component business cost indices that
measure the costs of labor, energy, and taxes relative to the U.S. national average.
CDBis = w_ulcis* ULCis / ULCiUS + w_eis * Energyis/EnergyiUS + w_t * Taxis/TaxiUS
where i = industry, s = state, US = U.S., w_ulc = unit labor cost weight, ULC = unit labor cost, w_e = energy weight, Energy = average electricity price, w_t = tax burden weight, Tax = effective business tax rate
An index value of 100 implies that the cost of doing business in industry i in state s is exactly
equal to the cost of doing business nationally.
The three components will not be used uniformly by each industry and in each state; the
weighting structure varies to account for these differences in cost structure. Moody's
Economy.com uses IMPLAN Professional's input-output accounts for 2007 to calculate the
proper weights for each component within in each industry and state.
The labor cost index measures unit labor costs in each industry within each state relative to the
U.S. For most industries, particularly in the service sector, labor costs comprise the largest share
of business costs. The variations across industries are wide, however. For example, in New York,
labor costs range from a low of 20 percent of the total Cost of Doing Business in the natural
resources and mining industry compared to a high of 98 percent in state government.
The energy cost index measures electricity prices (in cents per kilowatt-hour) for either
commercial or industrial electricity relative to the same for the nation. The price data come from
the Energy Information Administration. Manufacturing industries use industrial electricity and
most service sector industries use commercial electricity. Here again, there is a wide range across
industries within states. In New York, energy costs account for only 2 percent of total business
The Role of UI as An Automatic Appendix B-2 May 2010 Stabilizer During a Recession
costs in state government, and 67 percent in chemicals, energy, plastics and rubber
manufacturing.
An index of the state and local business tax burden is included to estimate the costs associate
with a state's taxes. A state's effective tax rate is measured as the total tax burden as a percent of
total personal income within an area, indexed to the national effective tax rate. Tax burden is
estimated using government revenues from taxes levied on personal property, corporate taxes,
and charges, less severance taxes. Corporate license taxes, education, hospital, and
intergovernmental transfers are included as well as business contributions to unemployment and
workers' compensation programs. Therefore, if a state generates more revenue through the
aforementioned taxes relative to incomes compared to the national average, its tax index will
exceed 100.